Effective application of process improvement patterns to business processes [original]

Softw Syst Model

DOI 10.1007/s10270-014-0443-z

SPECIAL SECTION PAPER

Effective application of process improvement patterns to business

processes

Matthias Lohrmann ·Manfred Reichert

Received: 24 September 2013 / Revised: 28 November 2014 / Accepted: 1 December 2014

Abstract Improving the operational effectiveness and effi-

ciency of processes is a fundamental task of business process

management (BPM). There exist many proposals of process

improvement patterns (PIPs) as practices that aim at support-

ing this goal. Selecting and implementing relevant PIPs are

therefore an important prerequisite for establishing process-

aware information systems in enterprises. Nevertheless, there

is still a gap regarding the validation of PIPs with respect to

their actual business value for a specific application scenario

before implementation investments are incurred. Based on

empirical research as well as experiences from BPM projects,

this paper proposes a method to tackle this challenge. Our

approach toward the assessment of process improvement pat-

terns considers real-world constraints such as the role of

senior stakeholders or the cost of adapting available IT sys-

tems. In addition, it outlines process improvement poten-

tials that arise from the information technology infrastruc-

ture available to organizations, particularly regarding the

combination of enterprise resource planning with business

process intelligence. Our approach is illustrated along a

real-world business process from human resource manage-

ment. The latter covers a transactional volume of about

29,000 process instances over a period of 1 year. Overall,

our approach enables both practitioners and researchers to

reasonably assess PIPs before taking any process implemen-

tation decision.

Communicated by Dr. Selmin Nurcan.

M. Lohrmann (B)·M. Reichert

Databases and Information Systems Institute, Ulm University,

Ulm, Germany

e-mail: [email protected]; [email protected]

M. Reichert

e-mail: [email protected]

Keywords Business process modeling ·Business process

design ·Business process optimization ·Business process

intelligence ·Business process quality

1 Introduction

Research on business process management (BPM) and

process-aware information systems (PAISs) has resulted

in many contributions that discuss options to improve the

quality, performance, and economic viability of business

processes [1]. Examples range from individual “best prac-

tices” [2] to comprehensive business process quality frame-

works [3,4]. In this context, we refer to process improvement

patterns (PIPs) as generic concepts for enhancing particular

aspects of business processes. As an example, consider deci-

sion processes that require to appraise various decision crite-

ria. The respective appraisal tasks can be arranged to reach a

decision with as little effort and as quickly as possible. This

can be achieved by executing tasks with a high probability

of providing sufficient information for a decision and with

comparably low execution effort earlier in the process. This

principle is known as “knockout” [5]. It constitutes a first

example of a process improvement pattern.

Example 1 (Knockout principle) Consider a process for han-

dling invoices received from suppliers. To determine whether

the invoice should be paid, we want to check whether it is in

line with purchase order data. In addition, we need to ensure

that there is a sign-off from the responsible manager. The

former check can be fully automated in the context of ERP

systems and therefore be executed with little effort. Thus, it

makes sense to execute this check first and possibly “knock

out” the invoice before incurring the much greater effort of

(manual) sign-off.

123

M. Lohrmann, M. Reichert

1.1 Research challenges

To ensure practical relevance, the actual business value of

PIPs needs to be demonstrated to practitioners, thus enabling

reasonable implementation decisions. In the context of this

issue, there exist many propositions for empirically establish-

ing the effectiveness of PIPs. These include anecdotal evi-

dence [6], case studies [7], and surveys [8]. Commonly, these

approaches are based on ex-post (i.e., hindsight) appraisal

of qualitative evidence given by process managers or other

stakeholders to obtain general insights applicable to compa-

rable cases.

However, there still exists a gap regarding the a priori

(i.e., in advance) assessment of PIPs considering a particu-

lar application scenario, which may range from an organiza-

tion’s strategy and goals to its existing business process and

information systems landscape. In particular, this gap should

be bridged for the following reasons:

– Similar to design patterns in software engineering [9],

PIPs constitute abstract concepts that may or may not be

useful in a particular context. Experience from other sce-

narios, which may widely differ from the one at hand, is

thus not sufficient to take reasonable decisions on the

implementation of organizational changes or process-

aware information systems (PAISs).

–Ex-post evidence is usually obtained from persons

involved in the respective implementation projects. In

turn, this leads to a source of bias. Moreover, a priori

assessment allows addressing a far wider spectrum of

PIPs. In particular, it is not necessary to complete imple-

mentation projects before a PIP can be assessed.

–Combining PAISs with process intelligence tools [10–12]

opens up new opportunities to quantitatively and qualita-

tively gauge real-world business processes. This should

be leveraged for scenario-specific PIP assessment.

–Effective PAIS development requires to consider process

improvement potentials beforeany implementation effort

is incurred. Accordingly, PAIS development should start

with a requirements definition which, in turn, is based on

adequate process design considering relevant PIPs.

To enable a priori PIP assessment, this paper tackles the

following challenges:

Challenge 1 Describe an approach toward a priori PIP

assessment reflecting and summarizing common practice in

the field.

Challenge 2 Evaluate the approach by applying it to a sub-

stantial real-world case.

Challenge 3 Reconcile the approach to scientific standards

by applying guidelines for empirical research in information

systems (IS).

1.2 Contribution

This article constitutes an amended version of the work we

presented in [13]. Reflecting the considerations made above,

[13] provides an approach toward a priori assessment of PIPs.

In particular, the contribution of [13] is based on the following

aspects:

– The proposal we presented in [13] provides a standard

approach to evaluate the impact of PIPs on organizational

objectives for specific application scenarios. Thus, it sup-

ports well-founded decisions on the implementation of

corresponding adaptations of business processes. More-

over, it contributes to bridge the gap between generic

PIPs proposed by the BPM research community and real-

world application scenarios. This way, it enhances the

practical relevance of proposed PIPs.

– It considers scientific rigor by applying an appropriate

research framework.

– It reflects practical requirements, which could be demon-

strated through an experience report covering a substan-

tial real-world business process.

In particular, when discussing the validity of our research

design, execution, and results with practitioners, we made

a number of observations that are generally applicable to

process improvement projects based on PIPs. Since these

may be helpful for practical application as well as future

research, they are included as project recommendations in

this paper.

This article complements our previous results with the

following additional contributions:

– It extends the presentation of the sample case used for our

experience report with empirical results based on process

mining [11]. Thus, it illustrates the application of this

technology to a practical case.

– It provides a discussion of open challenges regarding

process improvement tools.

– It provides a more profound discussion of the applied

process improvement methodology, thus rendering the

approach more accessible for application to other sce-

narios.

– It discusses the complete set of process improvement

measures that resulted from the application of PIPs to

the sample case.

– It describes the results obtained when revisiting our

process improvement measures 14months after having

completed the initial process improvement project.

– It provides a substantially extended discussion of related

work.

– It comprises a more detailed reflection of our results

against Challenges 1–3, as well as an assessment of the

123

Business process management

Fig. 1 Sample business

objective: handling incoming

applications

Job offer

sent

Letter of

refusal sent OR

approved

Agreement on

conditions

Symbols

Conditional

element

Subcondition:

relevant state of

conditional elements

AND

Necessary /

sufficient

subconditions

“Fully determinate” target aspect:

to be fulfilled if and only if

conditions are met

Technical

quality

Interview

results

AND

Senior

management

approval

Simplified terminology, full set of terms available in [15]

approved

given

not

approved

not

approved

not given

Fig. 2 Sample process: handling incoming applications (BPMN notation)

general applicability of our approach. This includes a dis-

cussion of relevant limitations and strategies to address

these.

The remainder of this article is structured as follows:

Sect. 2describes the sample process we use to illustrate our

approach. Section 3presents our approach toward PIP assess-

ment. In the sense of an experience report, Sect. 4describes

the results obtained when applying the approach to the sam-

ple process from Sect. 2. Section 5discusses the state of the

art in PIP assessment as well as other related work. Finally,

Sects. 6and 7evaluate our results referring to the challenges

discussed above and conclude the paper.

2 Sample case: applications management process

The business process we use to illustrate the concepts pre-

sented in this paper stems from the field of human resource

management. It addresses the handling of incoming job appli-

cations to fill open positions in a professional services firm.

Figure 1describes the business objective of this process

according to a notation we developed in [14]. The objec-

tive of the process is to achieve one of two states for each job

application: Either the application is refused, or a job offer is

sent to the applicant. A job offer shall be sent if the following

conditions are met: (1) The application documents have been

accepted in terms of quality (e.g., with regard to the CV), (2)

an interview has taken place with a positive feedback, (3)

basic conditions have been agreed on between both parties,

and (4) senior management approval has been obtained. If

one of these requirements is not met, a letter of refusal has

to be sent.

Based on our discussions with stakeholders and the results

of process mining, we can model the business process imple-

menting this business objective. For this purpose, we use

BPMN (cf. Fig. 2,[15]). For the sake of brevity, we slightly

simplify the model and omit a detailed description of its

elements. As an example of the relation between the busi-

ness objective and the process model, consider the condi-

tions the business objective poses toward sending a job offer.

The process model transforms these conditions into respec-

tive checking activities (e.g., Technical quality into Check

documents) and XOR decision gateways. Note that there is

not necessarily a one-on-one relation between conditions and

checking activities. Further, there may be multiple process

implementation alternatives for a given business objective

(e.g., multiple conditions may be checked within one activ-

ity).

123

M. Lohrmann, M. Reichert

Fig. 3 Termination states of the

application process: one fiscal

year sample

29.177

17.081

5.101

2.129 388 1.541 2.937

100% 59%

17%

7% 1% 5% 10%

Incoming

applications

Declined by

recruiting

department

Declined by

technical

department

Declined after

interview

Job offer not

approved

Withdrawal by

applicant

Applicant hired

Critical cases

Note: Analysis includes internships, but no back office applications (e.g.,

personal assistants); withdrawal by applicant can occur at any time

No. of process instances

Fig. 4 Filtered process map:

one fiscal year data sample

Figure 3breaks down the total number of applications

handled in a time period of one fiscal year into the num-

ber of applications for each possible termination state of

the process. Note that the termination states from Fig. 3

correspond to potential paths through the process model

from Fig. 2. We will refer to this overview when discussing

our research execution in Sect. 4. We obtained a corre-

sponding data sample of 27,205 process instances from the

log database tables of the PAIS supporting the business

process (in this case, an SAP ERP system). Each process

instance covers one application. Thus, 1,972 out of the 29,177

applications of Fig. 3are not included in the data sample.

These comprise, for example, applications handled in the

business units without involvement in the human resource

(HR) function. These applications are not traceable in the

PAIS.

Figure 4shows a process map generated with the Disco

process mining tool [16] when applying it to the data sample.1

For the sake of readability, this process map has been filtered

1The field of process intelligence deals with analyzing the actual enact-

ment of business processes [17]. In this context, process mining refers

to using processing events logged with a timestamp to generate process

maps, i.e., graphic representations of actual process enactment traces,

and additional process information [11]. Note that Disco was selected

as a representative of a number of tools available to practitioners in

commercial settings today. Alternatives like ProM [18]orCelonisDis-

covery [19] might be used as well.

123

Business process management

to solely comprise enactment traces that occur frequently and

events that are relevant for our analyses.

The process map is an example of the results that can be

generated with process mining tools. In the following, we will

use process mining and other techniques to analyze our log

data sample with respect to process improvement potentials.

The process map should be considered as an amendment, but

not as a replacement of “traditional” process models such as

the one presented in Fig. 2:

– The process map is based on events logged in the PAIS.

Not all events directly reflect a corresponding activity in

the process model, and identifiers of events might differ

from the ones of corresponding activities. There may be

activities not reflected in a logged event or events not

triggered by an activity from the process model.

– The process map shows the actual frequency of events in

the data sample. Thus, it reflects as-is process execution,

which may differ from to-be process design as recorded

in the process model.

– The process map needs to be interpreted with the sup-

port of experienced stakeholders. In our sample case, for

example, application refusal events are used to purge the

database of received applications to comply with privacy

regulations. Further, not all hirings are handled through

the corresponding end events. Issues like these need to be

understood when interpreting the process map. However,

this understanding is useful for process improvement as

well.

3 Methodology

Like other IS artifacts, PIPs constitute goal-bound artificial

constructs in the sense of the design science paradigm [20]

to be evaluated in terms of “value or utility” [21]. In our

context, this results in a particular challenge. While PIPs are

abstract concepts applicable to a broad range of scenarios,

their business value must be determined considering the spe-

cific use case to enable a decision whether the PIP should

be implemented. To this end, we use an extended conceptual

framework as summarized in Fig. 5.

Beyond the concepts of PIPs and business processes or

application scenarios, we introduce organizational objec-

tives, process improvement objectives, and process improve-

ment measures:

1. Organizational objectives reflect strategic goals an

organizationwantstoachievewithrespecttoanappli-

cation scenario. Examples of organizational objectives

that apply to many scenarios include the effectiveness of

process output, cost savings, or compliance with regu-

lations [1,22]. Note that these examples can be used as

Process improvement

objectives (PIOs)

Business process /

application scenario

Process improvement

measures (PIMs)

Organizational

objectives

Process improvement

patterns (PIPs)

SpecificGeneric

“What?”

“How?”

Additional concepts

Original concepts

Fig. 5 Extended conceptual framework

a starting point to identify organizational objectives rel-

evant to a particular application scenario. In principle,

such objectives are generic, but how they are prioritized

against each other is specific to an organization’s strategy.

2. Process improvement objectives (PIOs) comprise char-

acteristics that enhance a process considering orga-

nizational objectives. PIOs can be viewed as a refine-

ment of organizational objectives considering the par-

ticular challenges associated with a concrete application

scenario. In a step-by-step approach, PIOs can be refined

into a tree structure, as will be exemplified when dis-

cussing our application scenario in Sect. 4. The resulting

top-down model is a useful mental technique to ensure

acomprehensive perspective on process improvement.

Note that similar considerations are used in goal-oriented

requirements engineering (cf. Sect. 5.3) and value-based

management [23]. This procedure can be aborted as soon

as the resulting PIOs are sufficiently granular to allow for

the application of PIPs. PIOs thus constitute the “bridge”

between abstract organizational objectives and concrete

PIPs. The relevance of PIOs to organizational objectives

may be evident, or it may require additional validation.

As an example of immediately evident PIOs, consider the

elimination of obviously redundant tasks to reduce costs.

As an example of PIOs that require validation, consider

short cycle times. It is not necessarily a strategic goal to

enact processes as fast as possible. However, this may

be a PIO if a link between cycle times and a particu-

lar organizational objective (e.g., reducing costs) can be

demonstrated. PIOs thus provide an additional layer of

abstraction as a “shortcut” between improvement mea-

sures and organizational objectives. For the above exam-

ple, potential improvement measures might be validated

by demonstrating a positive impact on cycle times instead

of overall cost. PIOs can also be viewed as a tool to iden-

tify PIPs relevant for the application scenario: Available

PIPs are considered with regard to whether they can con-

tribute to a PIO. For example, the parallel execution of

formerly sequential tasks constitutes a PIP that may con-

tribute to shorter cycle times as an exemplary PIO. Note

123

M. Lohrmann, M. Reichert

Fig. 6 Problem statement and research design: required components

that the concept of PIOs corresponds to the identifica-

tion of stakeholders’ goals, which has been proposed as

a requirement for empirical IS research in [24].

3. Process improvement measures (PIMs) are bundles of

actions considered for joint implementation.2They

reflect the application of process improvement patterns

(PIPs) to a specific process in order to realize PIOs. Sev-

eral PIPs may be bundled into one PIM for joint imple-

mentation, depending on the given application scenario.

As an example of a PIM, consider the implementation

of a new workflow tool, which may incorporate multi-

ple abstract PIPs. A PIM thus applies one or more PIPs

to a specific business process to address one or more

particular PIOs. Assessing PIPs for a particular applica-

tion scenario thus amounts to the assessment of the busi-

ness value of corresponding PIMs considering relevant

PIOs.

Note that, considering the arrows, Fig. 5mayalsobe

read as a top-down method for process improvement. Sec-

tion 4further describes its application: General organiza-

tional objectives are refined to PIOs specific to the consid-

ered business process or application scenario. Then, PIPs rel-

evant to the concrete scenario are selected from a generic set

of generally available PIPs and bundled into concise PIMs.

Specifically to the application scenario, PIMs are described

in sufficient detail to enable to discuss and decide on their

implementation.

Business processes and PIMs, as our unit of study, are

implemented by means of PAISs. To maintain scientific

rigor, their assessment should take into account requirements

known from the empirical evaluation of propositions in soft-

ware engineering or IS research. In [24], the authors subsume

requirements in terms of scientific methodology for evalua-

tion approaches in IS research. Figure 6provides an overview

on the basic concepts described there. In the following, we

2Note that in the given context, the term “measure” is not to be under-

stood as a means of measuring something (e.g., a performance indicator)

or as a unit of quantity, but as a coordinated set of activities.

align our approach to [24]. We describe how each compo-

nent is reflected in our proposition. Note that the (general)

statements made should be further refined for each applica-

tion scenario. From a practical perspective, this will ensure

a common understanding by all project participants. Thus,

respective considerations are included in the following para-

graphs as well.

3.1 Problem statement

The first four components we address constitute the problem

statement according to [24].

Research question (“What do we want to know?”) Should

PIMs be implemented to better meet organizational objec-

tives? Note that this research question refers to PIMs instead

of PIPs in order to reflect our goal of scenario-specific assess-

ment.

For our sample case, the research question can be refined to

the question whether PIMs should be implemented to reduce

cost per hire (cf. Sect. 4.1).

Unit of study (“About what?”) The business process to be

improved and the proposed PIMs comprising PIPs constitute

our unit of study. Effectively selecting PIPs and bundling

them into scenario-specific PIMs require the participation of

knowledgeable, but also creative project members. For exam-

ple, the participants of workshops to discuss PIMs should be

carefully selected. In this regard, researchers may contribute

a valuable “outside-in view” based on, for example, experi-

ence from other scenarios.

Regarding our sample case, the application management

process and the proposed PIMs as the unit of study are

described in detail in Sects. 2and 4.3, respectively.

Relevantconcepts(“What do we know in advance?”) Related

work to be considered generally includes conceptual work on

PIPs, case studies on comparable processes, and benchmarks

available for the application scenario. In this regard, it is

helpful to ensure proper research of available literature as

well as a thorough use of available organizational knowledge

(e.g., through selection of appropriate interview partners).

For our exemplary application scenario, we use a frame-

work of process redesign practices [2], our own research into

PIPs, a cost per hire benchmark, and available research on

“knockout” processes [5].

Research goal (“Why do we want to know?”) Implementing

PIMs will result in cost and risks incurred (e.g., process dis-

ruptions). To avoid unnecessary cost and risks, implementa-

tion decisions should be based on appropriate investigation

of whether implementing PIMs will enable to better meet

organizational objectives. Implementation decisions should

consider not only benefits in day-to-day process operations,

but also required investments and future operating cost or

total cost of ownership.

123

Business process management

3.2 Research design

The five components described in the following constitute the

research design of our approach in terms of data collection,

measurement, and data analysis.

Unit of data collection To understand the application sce-

nario, we require an as-is process description to reflect

process design, a process instances sample to reflect process

execution, and PIOs to reflect refined organizational objec-

tives. Depending on the application scenario and practical

considerations, the process instances sample can be given

as a PAIS data extract, as a set of interviews with involved

people, as a set of cases directly observed, or as a combina-

tion thereof. To assess PIPs, we require descriptions of avail-

able PIPs and scenario-specific propositions of PIMs. Note

that data collection should cover both process design and

actual process execution. This way, PIOs can be identified

prospectively (based on the process design) and retrospec-

tively (based on the process execution). Immediate observa-

tions are preferable to indirectly related process information.

Depending on the application scenario and practical con-

siderations, the process instances sample can be given as a

PAIS data extract, as a set of interviews with involved peo-

ple, as a set of cases directly observed, or as a combination

thereof.

Regarding our sample application scenario, we could refer

to a business objective model and a flowchart of the process, a

statistic on the results of process execution, and a substantial

PAIS execution data extract (cf. Sect. 2). To assess PIPs, we

use PIP descriptions available in the literature and from our

own research, and PIMs as described below (cf. Sect. 4.3).

Environment of data collection Our proposition primarily

aims at improving existing business processes. Hence, data

are collected in the field to reflect the actual situation as best

as possible. The environment of data collection thus gen-

erally comprises process stakeholders (i.e., contact partners

involved in process execution, recipients of process output,

or suppliers of process input) as well as relevant documenta-

tion and PAISs. The environment of data collection should be

as broad as practically reasonable in order to facilitate iden-

tifying all PIOs that are relevant to organizational objectives,

and to enable appropriate assessment of PIPs and PIMs.

Regarding our sample case, the environment of data col-

lection comprised the head of recruiting, a business unit HR

partner, business unit team managers, the PAIS administra-

tor, and recruiting team members as process stakeholders. In

terms of documentation, we used regular recruiting manage-

ment reports and PAIS status codes. The PAIS used to support

the business process was available as well. As a limitation

to our sample environment of data collection, applicants as

a group of process stakeholders were not represented in the

environment of data collection due to practical requirements.

Because of privacy regulations, applicants’ contact data may

only be used to process the application, but not for other

purposes.

Measurement instruments Our approach is based on elabo-

rating PIOs and PIMs in a step-by-step approach. Draft PIOs

and PIMs are thus used to document input from the environ-

ment of data collection, and constitute measurement instru-

ments comparable to semi-structured questionnaires. These

are amended with the proceedings documentation from inter-

views and workshops (see measurement procedures below).

In addition, depending on the process instances sample,

process execution tracing capabilities in PAISs or statistical

process control (SPC) procedures also need to be considered.

Note that measurement instruments should consider usability

criteria with regard to stakeholders involved in measurement

procedures. For example, this requires using terms customary

to the organization when phrasing PIOs and PIMs.

Regarding our sample application scenario, PIOs and

PIMs used as measurement instruments are described in

Sects. 4.2 and 4.3, respectively. In addition, we used work-

shop proceedings, confirmation letters on results reconcili-

ation (via email), and procedures to extract execution data

from the PAIS used to manage incoming applications.

Measurement procedures Depending on the application sce-

nario and practical considerations, relevant measurement

procedures comprise stakeholder interviews, stakeholder

workshops, and questionnaire procedures. Process mining

can be used if the sample of process instances is based on

a PAIS data extract. Measurement procedures should take

into account customary practices of the organization, e.g.,

by using standard templates for meeting proceedings.

On-site measurement procedures (i.e., observing the

process in its operations environment) can help to identify

additional PIOs to be addressed for process improvement by

giving a clearer picture of day-to-day process issues.

Regarding our sample case, we used telephone and face-

to-face interviews with follow-up reconciliation of proceed-

ings, a recruitment center site visit, and process mining with

Disco.

Data analysis procedures In general, relevant data analy-

sis procedures include qualitative analysis of workshop

and interview results and quantitative analyses of process

instance samples depending on the measurement instruments

applied. Note that data analysis procedures need to be flexi-

bly adapted to the step-by-step refinement of PIOs and PIMs

and to the form of quantitative data available on the process

instances sample. In practice, this may lead to a mix of

tools actually applied. In this context, for example, statistical

analysis tools can significantly reduce quantitative analysis

effort and therefore enable enhancing the search scope for

relevant PIOs.

Regarding our sample case, a qualitative analysis was con-

ducted together with stakeholders as described in Sect. 4.3.

In turn, the quantitative analysis comprised filtering of sub-

123

M. Lohrmann, M. Reichert

Fig. 7 Deriving process

improvement objectives

process views in a process mining tool (Disco), re-extraction

of filtered samples and import into a spreadsheet application,

conversion of the event log into a “case log” (i.e., an array of

events for each process instance), computation of cycle time

attributes for each case, and statistical analysis with Minitab.

4 Sample case: process improvement patterns

assessment

We now apply our extended conceptual framework com-

prising organizational objectives, PIOs, PIPs, and PIMs (cf.

Fig. 5) as well as our research design to our sample appli-

cation scenario. Further, we summarize our observations

regarding the use of tools and systems for empirically ana-

lyzing our sample process, which may be useful for further

developments in this regard.

4.1 Organizational objectives

As discussed, obtaining clarity about the content and business

value of organizational objectives constitutes a fundamental

prerequisite to ensure the relevance of PIP assessment. In our

sample application field (i.e., recruiting), organizations strive

to fill vacant positions quickly, cost-effectively, and with suit-

able candidates. To achieve these goals, personnel market-

ing is responsible to generate a sufficient number of suitable

applications, while the purpose of our sample process (i.e.,

managing job applications) is to convert applications into

actual hires.

Thus, organizational objectives for the sample applica-

tion scenario include reducing the time needed until open

positions are filled,reducing cost per hire, and improving

the quality of hired applicants. Out of this set of objec-

tives, reducing cost per hire is well suited for illustrating

our approach. In particular, the issue of cost is transferable

to many other scenarios. More precisely, the following con-

siderations apply for our sample process:

Reducing cost per hire as organizational objective The

cost per hire key performance indicator captures the total

cost of both personnel marketing and applications manage-

ment. While recruiting cost spent per application is propri-

etary data, based on experiences from projects with clients

we assume an amount of about 400 Euros. In our sam-

ple scenario, generating and managing about 29,000 appli-

cations per year would thus result in 11.6m Euros total

cost, with cost per hire at around 4,000 Euros. Since hir-

ing cost for talent in professional services will be higher

than in, for example, manufacturing, this value corresponds

well to the average of 4,285 USD reported as cost per hire

for larger organizations by a benchmarking organization

[25]. Further, it seems rather conservative considering that

professional recruiting consultants commonly charge half

a year’s salary for successful hires, depending on industry.

This calculation demonstrates the high relevance of reduc-

ing cost per hire through an improved application handling

process.

Note that while reducing cost per hire has been chosen

to illustrate our approach, the other objectives remain highly

relevant. In particular, they need to be kept in mind when

designing PIMs to avoid improving the process toward one

objective at the expense of others. As an example, improving

recruitment cost should not result in eliminating face-to-face

interviews with candidates since this would probably reduce

cost at the expense of the quality of applicants hired.

4.2 Process improvement objectives (PIOs)

PIOs pertain to characteristics of the business process that

affect the organizational objectives we want to improve on.

They serve as a “shortcut” to facilitate discussing the business

value of PIMs without reverting to high-level organizational

objectives. In our sample case, we refine the organizational

objective to reducecostperhirein a step-by-step approach by

asking the question what drives cost per hire or, subsequently,

the resulting lower-level PIOs.

Figure 7presents relevant aspects of the resulting tree

structure: Initially, total cost per hire is driven by the cost

associated with each process instance (i.e., with each individ-

123

Business process management

ual application), and, since it is possible that multiple process

instances are needed to fill one position, with the overall

number of process instances required. Both aspects may be

optimized to reduce cost per hire, but are still rather abstract

and will not allow applying PIPs without further refinement.

On the one hand, cost per process instance is determined

by the cost of production factors (e.g., the cost of employees’

working time or the cost of IT systems used) and the amount

of effort spent on each process instance. Both drivers will

occur in any tree of PIOs dealing with cost reduction. Fac-

tor costs, however, are generally unsuitable as a PIO since

they are not governed by process designers and managers.

Therefore, they cannot be subjected to process improvement

efforts. Rather, they should be considered as a factor given

externally which may affect assessment results. As an exam-

ple, consider the impact of location decisions on labor costs.

Example 2 (The impact of factor costs on PIP evaluation)

Particularly in large organizations, it is a common practice

to bundle administrative business processes into “shared ser-

vices organizations” [26]. In this context, labor costs con-

stitute an important factor when deciding on the location

of the shared services organization. In turn, this decision

impacts considerations on the economic viability of process

improvement measures. For example, when considering cap-

ital investments to automate manual activities, like match-

ing incoming payments on bank statements against invoices

issued to customers, lower labor costs will increase the pay-

back time of the investment, thus rendering its implementa-

tion less attractive.

On the other hand, cost per process instance is determined

by the quantity of production factors associated with each

process instance. In our sample process, factors besides man-

ual labor can be neglected, as will be the case for most admin-

istrative business processes. Accordingly, reducing effort per

process instance pertains to reducing manual processing

effort. This PIO lies at the core of many PIPs commonly

used in practice, such as task elimination,task automation,

or knockout [2], and has thus reached a sufficient degree of

refinement.

Besides reducing cost per process instance,cost per

hire might be improved by reducing the number of process

instances required over time. This option corresponds to the

elimination of in-efficacious process instances that do not

terminate in a desirable state according to the underlying

business objective. It closely resembles methods applied in

common quality management approaches that aim at reduc-

ing “causes of poor quality” [27]. In particular, every in-

efficacious process instance can be viewed as a quality issue

in the business process. Note that the overall effect of qual-

ity management on cost and firm performance has been

well recognized and empirically demonstrated [28]. This

option is particularly interesting since associated measures

can often be implemented with limited investments. Hence,

further considerations on our sample case will focus on this

PIO.

In our sample case, cost per hire is driven by the general

efficiency of the application management process, but also

by the frequency of process instances terminating in one of

the states marked as “critical” in Fig. 3. The following con-

siderations apply in this regard:

–Not approving a job offer after a successful interview

may be caused by defective steering of capacities (i.e.,

job vacancies), defective communication of terms to be

offered, or defective review of application documents.

–Job offers declined by applicants mostly means that the

applicant does not approve of conditions offered, did not

have a good impression during the application process,

or has decided to take another job offer.

Since terminating the process in these states means that

significant effort has been incurred while still failing to hire

a promising candidate, organizational objectives are clearly

violated: On average, only one out of six applications will

successfully pass interviews. However, considering critical

cases with defective termination events (cf. Fig. 3), only one

out of ten applicants can be hired. In other words, if the

process enactment defects lined out could be fully eliminated,

only about 18,000 applications would have to be acquired and

managed to cover demand. This would reduce total hiring

cost by about 4.6m Euros.

Based on the considerations made above, we focus on

PIOs to reduce effort per process instance or to reduce the

probability of the defective process outcomes described.

Table 1summarizes the resulting topics.

Note that for the first PIO (i.e., Reduce manual processing

effort) there is an evident link to our organizational objective

of reducing cost per hire. However, the second and third

PIOs (i.e., Reduce failed approvals and Reduce cycle times)

are based on hypotheses on what can be done to reduce

process enactment defects affecting the organizational objec-

tive. Accordingly, they require qualitative or quantitative evi-

dence to corroborate their relevance for reducing defects and

thus improving cost per hire.

For the second PIO (i.e., Reduce failed approvals), we

obtained qualitative evidence by interviewing responsible

managers, which confirmed the topics described in Table 1.

Since the reasons for failed approvals are not captured in the

PAIS used to manage the application process, quantitative

evidence is not available.

For the third PIO (i.e., Reduce cycle times), the causal

link to its underlying defect of applications withdrawn by

candidates is not as obvious. Further quantitative analysis

is thus required. Figure 8summarizes the duration between

interviews and job offers for the subsets of applicants accept-

123

M. Lohrmann, M. Reichert

Table 1 Sample case: process

improvement objectives PIOs Rationale

Reduce manual processing effort Emerging potentials in terms of reducing process enactment effort

per instance should be addressed

Reduce failed approvals Final approval of job offers by senior management fails if there are

issues regarding vacancy management, reconciliation of terms, or

checking of documents. The probability of these “late defects”

should be addressed

Reduce cycle times The probability of applicants’ obtaining and taking alternative job

offers increases with time. Therefore, cycle times between

applications being received and job offers made should be as

short as possible

Fig. 8 Boxplot Duration interview to job offer in weeks vs. acceptance

of job offers

ing and declining their offer in a boxplot (this part of total

cycle time will later be the subject of process improvement,

cf. Sect. 4.3). In the figure, the differences between subsets

regarding quartiles, median, and mean values appear as rela-

tively small. However, a correlation between cycle times and

the probability of a candidate to accept or decline a job offer

can be statistically demonstrated

Correlation between job offers declined and cycle times

We want to determine whether there is a significant influence

of cycle time between application receipt and job offer in

weeks on the probability of an applicant accepting or declin-

ing a job offer. Accordingly, we use a binary logistic regres-

sion test to evaluate the influence of a metric independent

variable on a binary dependent variable. For this test, we use

a sample of 2,721 job offers representing about 70 % of the

annual volume (cf. Fig. 3) and consisting of instances fully

covered in the PAIS (not all interviews and feedbacks are

documented in the PAIS). The sample contains 261 cases

where the job offer was eventually declined by the applicant.

This is the latest point in the process where withdrawal by the

applicant is possible, and a significant amount of effort will

have been spent on each respective case. The two samples

are independent from each other and have a size of more than

100 cases each. Thus, the binary logistic regression can be

applied.

Figure 9shows an excerpt from the output of the statisti-

cal software package we used (Minitab). The p-value of less

than 0.05 indicates sufficient evidence to assume a signifi-

cant correlation between the occurrence of withdrawal and

cycle time. Regarding the question whether this correlation

can be interpreted as a causal link of cycle times impacting

the probability of withdrawal, the following considerations

apply:

– Cycle time is measured between receipt of the application

and the ultimate feedback to the candidate, whereas the

withdrawal sample refers to candidates that declined a

job offer thereafter. An impact of the occurrence of a

withdrawal on cycle time can therefore be ruled out.

– There is a plausible explanation for longer cycle times

causing withdrawals: It is possible that candidates find

another job while waiting for feedback after an interview.

This explanation is substantiated by data on withdrawal

reasons collected for a sample of 51 withdrawals between

October 2013 and January 2014 for one business unit. The

sample covers only cases where a reason was given for the

withdrawal. In 33 out of 51 cases, the reason cited was a

job offer by a third party, and we may assume that longer

cycle times provide more opportunities for candidates to

find alternative employment.

Fig. 9 Minitab output excerpt:

binary regression test Logistic Regression Table

Odds 95% CI

Predictor Coef SE Coef Z P Ratio Lower Upper

Constant -2.58986 0.169500 -15.28 0.000

Duration_weeks 0.144378 0.0635831 2.27 0.023 1.16 1.02 1.31

P-Value: Probability

of duration not being a

relevant factor

Odds Ratio: Lowering

duration by one week

expected to reduce

withdrawal risk by 16%

123

Business process management

Table 2 Defining process improvement measures for process improvement objectives

PIOs Applicable PIMs with comprised PIPs

Reduce manual processing effort PIM 1 (Application management automation): Task automationa, rout-

ing automation

Reduce failed approvals PIM 2 (Utilization and capacity management): Empowermenta,

knockouta,b

PIM 3 (Standardized terms and conditions): Triagea, bufferinga

Reduce cycle times PIM 4 (Managing interview feedback cycle times for successful appli-

cants): Control additiona, routing automation, escalation procedure

PIM5(Improving application routing): Case managera, knockouta,mit-

igation of repetitive loops

aProcess redesign practices comprised in [2]bAccording to [2], knockout decisions should be ordered “in a decreasing order of effort and in an

increasing order of termination probability”. Since this would contradict the goal to knock out respective instances with as little effort as possible,

we assume that knockout decisions should be arranged in reversed order

– We discussed potential additional independent variables

with a positive effect on both cycle times and the proba-

bility of withdrawal with HR management and obtained

no evidence on such variables. HR managers even men-

tioned that particularly sought-after candidates, who can

be expected to quickly obtain alternative job offers, are

handled with higher priority by business units. This effect

might even “hide” part of the correlation between cycle

times and probability of withdrawal. However, quantita-

tive evidence on this issue is not available.

According to the “odds ratio” column, a 1-week delay can

thus be expected to increase the probability of an applicant

declining a job offer by 16 %.

The significant correlation between cycle times and the

probability of withdrawal did not become obvious when just

considering median and mean values, but only when execut-

ing the binary logic regression test. This observation stresses

the necessity to use both sufficient sample sizes and appro-

priate statistical methods when dealing with empirical data

on business process enactment.

4.3 Process improvement measures (PIMs)

PIPs relevant to our application scenario have been selected

by considering a “longlist” of known PIPs in terms of poten-

tial contributions to the PIOs described above. In our case,

the “longlist” comprised PIPs from a framework by Rei-

jers and Limam Mansar on process redesign practices [2]

(these are marked with an asterisk “*”) as well as from

our ongoing research on improving business process quality

[4,29]. However, organizations are not limited with regard

to the sources of PIPs that can be used. PIOs are thus

used as a mental technique to guide the identification of

patterns that are useful for the organization. Relevant PIPs

are then bundled into PIMs specific to the application sce-

nario.

Table 2summarizes PIOs and corresponding PIMs as bun-

dles of PIPs.

In actual design and implementation projects, it is com-

mon to document and track individual PIMs through mea-

sure cards. In the following, we describe the PIMs from

Table 2in more detail using this metaphor. For each PIM,

we give a short content description—with PIPs involved

marked as italic—and additional details on the following

issues:

–Implementation describes steps required to realize the

measure.

–Business value appraises the expected implications con-

sidering the impact on PIOs as well as implementation

effort.

–Stakeholder verification describes the results of vali-

dating the PIM through interviews with respective stake-

holders.

Note that the PIMs presented here are, by definition, spe-

cific to our application scenario. However, their structure as

well as the underlying methodology is generally applica-

ble. In terms of content, they exemplify issues commonly

found in business process improvement projects, such as the

evaluation of IT implementation effort. Moreover, beyond

the scope presented here, actual measure cards comprise

additional information relevant to project management. This

includes project planning, project organization, key mile-

stones with “traffic light” status, risks, next steps, and deci-

sion requirements. Reporting on measure cards usually takes

place in steering committee meetings of senior manage-

ment.

PIM Card 1 (Applicationmanagementautomation)Our sam-

ple process contains manual activities which might be automated

using the PAIS. This pertains to the Assign contact partner and

Plan interview tasks enacted by the recruiting function. According

123

M. Lohrmann, M. Reichert

to the recruiting statistics presented in Fig. 3, these tasks occurred

in 12,096 and 6,995 process instances, respectively. In this regard,

further automation could be achieved by using master data already

available in the PAIS.

Implementation. The manual assignment of an interview partner

might be eliminated by implementing routing automation instead.

Each process instance in the PAIS is assigned to a job adver-

tisement. This is done either by the candidate when entering his

application data via a web platform provided by an external ser-

vice provider, or by the recruiting function. Since each job ad

refers to a set of business unit contact partners, these data might

be used to implement routing automation.However,automating

the assignment of contact partners requires implementing tight

control over master data quality to ensure that appropriate con-

tact partners are maintained for all job ads. Currently, this topic

remains challenging due to the quantity of job ads, which are

specific to service, location, degree of job experience, and other

factors.

In addition, the Plan interview task might be automated by

replacing the verbal feedback on candidates given by business

units with structured data including the relevant interview partner.

In this case, however, each job ad should be assigned to only one

business unit contact partner in order to avoid the need of coordi-

nating feedback from multiple sources.

Business value. The manual effort involved in routing an applica-

tion to a contact partner is estimated to be about 5 minutes by the

head of recruiting. Accordingly, about 90 routings can be handled

in one working day, resulting in a capacity reduction potential

of 0.7 full-time equivalents (FTEs) for about 12,000 routings per

year. This would amount to about 49,000 Euros annual cost sav-

ings. Implementation of automated routing (including the required

function of defining deputies for contact partners) is estimated to

require about 20 consultant days at a cost of 21,800 Euros. In

addition, the necessary master data cleanup of currently about

400 job ads is expected to take about 150 person days includ-

ing project setup and reconciliation with business units, corre-

sponding to 52,500 Euros one-off cost. Moreover, implementing

a workflow to ensure future master data quality when defining job

ads is expected to require 35 consultant days at a cost of 38,150

Euros.

Implementing automated interview planning builds on the

improved quality of contact partner master data and would require

an additional 10 consultant days or 10,900 Euros to implement the

required structured interview partner feedback. Interview coordi-

nation is currently estimated to take about 15 minutes per process

instance, corresponding to 1.1 FTEs or 77,000 Euros annual cost.

In total, recurring annual cost savings of 126,000 Euros are

to be matched against one-off implementation cost of 123,350

Euros.

Stakeholder verification. When discussing this measure with

stakeholders, the required change to the assignment of contact

partners to job ads proved as the most challenging issue. In par-

ticular, the definition of location-specific job ads raises concerns

with stakeholders that this might lead to candidates applying for

multiple job ads simultaneously. With regard to the current policy

of defining job ads, the implementation decision on this measure

card was postponed to late 2013, because the organization plans

to re-examine its entire set of job ads with the goal of substantial

reduction. Stakeholders assume that this will improve the “cus-

tomer experience” of candidates. However, it would require to

retain manual assignment of contact partners based on applica-

tion documents. Note that, counterintuitively, in this case sim-

plification of master data would thus lead to increased manual

effort.

PIM Card 2 addresses the reduction in failed approvals as

one of the critical cases identified in Fig. 3.

PIM Card 2 (Utilization and capacity management) Among

other reasons, senior managers refuse to approve job offers when

the business unit that wants to hire a candidate has excess capacity.

In our case, this can be traced by monitoring personnel utiliza-

tion, i.e., the rate of hours booked on client projects. If utilization

falls below a certain level, there are not enough client projects

for present staff, i.e., there is excess capacity. While refusal rea-

sons are not tracked in the PAIS, stakeholder interviews resulted

in an estimate of about 30% of total refusals to be caused by

this issue. Since candidates’ qualifications, in particular in gradu-

ate recruiting, are mostly not specific to particular business units,

the recruiting department can be empowered to route applications

to more appropriate teams from the start on. This results in an

early knock out of applications that would, in the end, be declined

because of low utilization.

Implementation. To enable utilization-based routing decisions,

a new report on utilization per team must be integrated into the

application management workflow. Since relevant data are avail-

able and are routinely retrieved for other reporting processes,

implementation effort has been estimated to be 25 consultant days

or 27,500 Euros. In addition, relevant utilization thresholds must

be agreed and communicated. The recruiting center routes about

12,000 applications per year. If the additional operating effort for

the utilization check can be assumed to be 10 minutes per appli-

cation, this results in an overall additional capacity requirement

of about 1.2 full-time equivalents (FTEs), amounting to approxi-

mately 84,000 Euros annual cost.

Business value. The PIM is expected to reduce the “late refusal”

rate by about 30% or 120 cases per year. Assuming a rate of job

offers declined by the applicant of 7 % (cf. Sect. 4.2), this would

reduce the number of applications to be generated and managed

to achieve a constant volume of hires by about 1,200 per year.

As we assumed the cost per application to be about 400 Euros,

an annual savings potential of 480,000 Euros compares to 27,500

Euros one-off cost and 84,000 Euros operating expenditure per

year.

Stakeholder verification. When discussing the PIM with senior

stakeholders, its business value appeared as rather clear. How-

ever, the distribution of utilization data emerged as a “political”

issue. Considering present organizational culture, the PIM will

not be implemented right away, but the basic capability to add uti-

lization control functionality to the PAIS will be included with the

requirements definition for the new PAIS solution to be completed

by early 2013.

The abbreviated measure cards presented above exem-

plify how PIM implementation benefits can be projected and

matched against expected implementation effort. However,

beyond this quantitative reasoning, qualitative (or “political”)

topics may play a role in taking implementation decisions as

well, as will be exemplified with PIM Cards 3 and 4.

PIM Card 3 (Standardized terms and conditions) In our sam-

ple process, extending contract offers to successful candidates

requires final senior management approval. This pertains not only

to the candidate, but also to the terms and conditions offered. In

123

Business process management

this regard, salaries are of particular importance, since this topic

is often discussed with the candidate during the interview or in a

follow-up conversation. In some cases, this leads to senior man-

agement approval not being granted because of terms that have

been offered to candidates in verbal agreements. Again, refusal

reasons are not tracked in the PAIS, but this issue is estimated to

cause about 25% of refusals. To remedy this topic, binding stan-

dard terms and conditions could be made available to business

unit interview partners. This would implement the information

buffering PIP since information is provided upfront and does not

need to be actively obtained by contact partners. In addition, an

additional process step of discussing potential terms in advance

between HR and the business unit might be included for partic-

ularly relevant candidates (e.g., applicants with job experience).

This would implement the triage PIP since it would include an

additional task for some cases.

Implementation. Implementation of this PIM might be achieved

by organizational measures without changes to the underlying

PAIS, since the upfront discussion of potential conditions can be

included in the existing workflow between HR and the business

unit. Figure 10 presents the corresponding change to the underly-

ing process model.

Note that without implementing additional workflow activities

or status codes, this change could not be ascertained by process

mining. The number of interviews on manager level is available in

the data sample and makes up 5.9% of total volume. Contact part-

ners estimate that about the same relevant volume will be caused

by other candidates. Accordingly, approximately 3,200 discus-

sions on terms would have to be conducted annually. Assuming

that about 10 discussions can be handled in one working day by

HR, this results in an additional required capacity of 1.8 FTEs at

about 126,000 Euros cost per year. Additional effort on part of

the business unit is not considered. Finally, the cost of compil-

ing and communicating standardized terms and conditions can be

neglected.

Business value. The PIM is expected to reduce the “late refusal”

rate by about 25% or 100 cases per year, with an annual savings

potential of 400,000 Euros according to the calculation presented

with PIM 2, in comparison with 126,000 Euros additional operat-

ing expenditure per year.

Stakeholder verification. The proposition of fully standardized

terms and conditions raised concerns with stakeholders that this

loss in flexibility might negatively impact chances to win candi-

dates. However, due to privacy regulations, no empirical evidence

regarding the impact of terms and conditions (namely, salaries)

on the probability of candidates to accept contract offers could

be obtained. Therefore, the decision to implement this PIM was

postponed until conclusions from an internal confidential assess-

ment are available, which will be additionally reconciled with an

HR consultancy.

PIM Card 4 addresses the Reduce cycle times PIO by deal-

ing with one of the underlying drivers for unnecessarily long

cycle times.

PIM Card 4 (Managing interview feedback cycle times for

successful applicants) The time span between successful inter-

views and job offers can be reduced by implementing a control

addition. This means that additional control flow elements are

included to ensure the correct enactment of the process. Triggered

through routing automation, the recruiting department will call

the interviewer directly when feedback is not available five busi-

ness days after an interview. If the interviewer cannot be reached

within two business days, an escalation procedure will take place

by calling the respective supervisor. If no feedback can be obtained

through these PIPs within ten business days, a letter of refusal will

be sent.

Implementation. To implement the PIM, comprehensive tracing

of interview dates and an additional workflow with corresponding

triggering mechanisms must be implemented in the PAIS. This

results in an estimated cost of approximately 38,500 Euros for 35

consultant days. In the data sample used for the binary regres-

sion test (cf. Sect. 4.2), about 51% of cases would fall under the

proposed regulations. Thus, a total volume of 7,000 interviews

conducted annually (cf. Fig. 3) would result in about 3,600 esca-

lation procedures. On the one hand, this number can be expected

to decline over time. On the other hand, multiple phone calls might

become necessary for one escalation case. Hence, we assume that

20 escalation procedures can be handled in one person day, i.e., an

additional 0.9 FTEs are required, resulting in about 63,000 Euros

annual cost.

Business value. Based on our binary logical regression analysis

(cf. Sect. 4.2), we reconciled with stakeholders that the maxi-

mum interview feedback time can be reduced to 2 weeks based

on an escalation process. Applying the corresponding odds ratio

(cf. Sect. 4.2) to all cases in our sample exceeding this time-

frame results in a reduction of 39.2 cases of “late withdrawals”

(cf. Fig. 3). This would reduce the number of applications to be

generated and managed by about 390 per year, corresponding to

156,000 Euros in annual savings. Considering additional operat-

ing expenditures of 63,000 Euros results in a total annual cost

reduction of 93,000 Euros versus a one-off cost of 38,500 Euros.

Stakeholder verification. During stakeholder interviews, we val-

idated implementation cost with the application workflow admin-

istrator, additional processing effort at the recruiting center with

the head of recruiting, and overall viability of the new process with

the head of recruiting and the business unit HR partner. The esca-

lation procedure to provide timely feedback was challenged by the

business unit HR partner, but not by team managers. Final consent

on the positive business value of the PIM could be achieved by

discussing the quantitative analysis of the underlying PIO.

The final PIM we identified exemplifies an issue that

occurs regularly in a top-down approach as employed in our

case: Since it is possible that similar PIPs can be used to

address various PIOs, overlaps in PIMs content may emerge.

This topic needs to be considered in implementation planning

and when consolidating recommended PIMs into a “manage-

ment summary” view (e.g., in terms of overall implementa-

tion cost and cost savings potentials). In general, it is prefer-

able to consolidate corresponding PIMs into one overall PIM

addressing multiple PIOs. However, even in this case, the top-

down approach facilitates obtaining a clear overall picture of

potentials to be realized by PIP implementation.

PIM Card 5 (Improving application routing) Similar to PIM

1, Application Management Automation,thisPIMdealswith

the implications arising from defects in application routing.

123

M. Lohrmann, M. Reichert

Fig. 10 Changes to the sample process when implementing PIM 3

Besides additional manual effort, loops in application rout-

ing, i.e., forwarding applications from one contact partner to

another, extend cycle times. Unfortunately, the practice of for-

warding application data is currently not tracked in the ana-

lyzed PAIS since documents are mostly forwarded manually

via email. Accordingly, implications of this practice cannot be

quantified. This issue underlines an additional benefit of com-

prehensive implementation and use of PAISs: achieving trans-

parency over actual process enactment, and thus improvement

potentials.

Implementation. In terms of implementation, this PIM closely

corresponds to PIM 1, Application Management Automation,

since it requires improving the assignment of business unit contact

partners through appropriate master data. Again, implementation

necessitates defining one case manager contact partner for each

job advertisement. Job ads would have to be defined in a man-

ner sufficiently granular to enable this 1:n relation between ads

and contact partners, specifying service, grade of job experience,

and location. Thus, implementation would enable us to mitigate

repetitive loops by assigning the right contact partner from the

beginning on and to achieve early knockout of unsuitable candi-

dates.

Business value. Based on the available empirical data set, the

business value of implementing this PIM cannot be quantified

since repetitive loops are not consistently tracked in the analyzed

PAIS. To enable tracking, the manual forwarding of application

data would have to be replaced by a corresponding PAIS function.

Once this function has been implemented, it will be possible to

assess the impact of loops on cycle times through statistical test-

ing of the impact of the number of loop iterations on overall cycle

time.

Stakeholder verification. For the purpose of stakeholder discus-

sion, the content of this PIM was merged with PIM 1, Applica-

tionManagementAutomation. Accordingly, the same implications

apply, namely to postpone final implementation decision until the

overall restructuring of job ads planned for late 2013. It is expected

that the implementation of a new application management PAIS

will facilitate to quantify the impact of repetitive loops on cycle

times and, ultimately, candidates’ probability of withdrawing their

application by then.

Figure 11 compares PIMs 1–4 in terms of one-off imple-

mentation cost and recurring savings potential per year (i.e.,

gross cost reduction minus additional operating effort, PIM

5 could not be quantified in this respect). Note that the pre-

sented PIMs exhibit a fairly positive business case, with

ratios between implementation cost and total annual sav-

ings below 1 year, and that the most positive business cases

can be realized by implementing organizational measures

without expensive IT implementation. They constitute good

examples of a phenomenon often encountered in practice:

in many cases, it is interesting to first identify and resolve

existing process defects within the framework of available

technology before additional process automation is imple-

mented at huge cost. Once these “quick win” potentials

have been leveraged, further process automation should be

considered.

4.4 Implementation results

The process improvement project facilitating our sample case

has been concluded in early 2013 with the go-live of the

newly implemented PAIS. This section briefly revisits the

PIMs discussed above with regard to the results actually

achieved. Statements are based on follow-up interviews con-

ducted with stakeholders in March 2014, i.e., about 1year

after go-live. Our interview partners included the head of

recruiting operations, the application management process

manager, the HR partner of a business unit, and two business

unit team managers involved in recruiting (e.g., as interview-

ers of applicants). To structure the interviews, we used the

available PIM cards and collected feedback on their imple-

mentation and corresponding results.

PIM Card 1 (application management automation) Dis-

cussing this measure with stakeholders during our analysis

phase resulted in postponement of the implementation deci-

sion because of the required restructuring of job ads. By now,

it has been decided to implement the PIM with slight changes

as discussed in the following. This decision has been taken

because the demonstration of business value documented

in PIM Card 1 has led to increased management attention

regarding the avoidable manual effort spent on routing appli-

cations. The organization is currently undergoing an effort to

significantly reduce variability in job ads. In the future, each

123

Business process management

Fig. 11 Comparison of PIMs

job ad will have exactly one contact partner from a business

unit assigned who will automatically receive screened appli-

cations. If the contact partner wants to pass the application to

her colleagues for an interview, she will choose the appropri-

ate person from a contact partner data base. This way, manual

routing of applications can be largely eliminated.

PIM Card 2 (utilization and capacity management) The

issue of utilization and capacity management has been

referred to an entirely new “workforce management system”

currently under development. This system will interface with

the application management PAIS to avoid the issue of rout-

ing applications to teams with limited utilization. Note that

this functionality will build on the implementation of PIM 1

as discussed above by controlling proposed contact partners

from the contact partners database.

PIM Card 3 (standardized terms and conditions) Terms

and conditions offered to candidates have been reconciled

with an HR consultancy. This assessment has led to the result

that current terms and conditions are in line with applicable

benchmark values. On that basis, a new policy has been issued

that requires deviating terms to be reconciled with business

unit management. According to our interview partners, this

policy has reduced corresponding cases to a minimum, which

has led to a significant reduction in “late refusals” as proposed

in the PIM.

PIM Card 4 (managing interview feedback cycle times

for successful applicants) According to application man-

agement reporting, interview feedback cycle times could be

reduced to an average value of 1.4weeks based on imple-

menting this PIM. Concurrently, the share of “late with-

drawals” could be reduced to 7.7% for the timespan of Octo-

ber 2013 to March 2014, in comparison with 9.6 % in the

original data sample we analyzed. However, one needs to

keep in mind that this reduction might be caused by varying

reasons, such as changes in the labor market environment.

Nevertheless, our interview partners confirmed their impres-

sion that reducing cycle times significantly contributed to this

development.

PIM Card 5 (improving application routing) As proposed,

this measure is being implemented in conjunction with PIM

Card 1, Application Management Automation, namely in

relation to managing job ads master data. Extensive loops and

cycle times during application are now controlled by main-

taining the responsibility of initial contact partners for timely

feedback, even if the application is passed on to colleagues

within the business unit. This way, contact partners have an

incentive to avoid redundant loops. Outstanding feedback is

then escalated to senior management.

4.5 Deployment of tools for empirical process analysis in

practice

This section amends the experience report on our sample

case with practical challenges observed when using available

technology to drive the empirical analysis of process data.

We believe that our sample scenario is fairly typical in this

regard and that the issues described may be useful for the

further development of corresponding tools and systems. In

our empirical analysis, we used three types of technology:

first, a process-aware ERP system, second, a process mining

tool, and third, a tool for statistical analysis.

When using the ERP system of our sample process as a

source of information for process improvement, we found

it rather challenging to extract and interpret empirical data

on process enactment. The major issues in this respect lay

in the complexity of the underlying data model, which was

partly subject to customization, its available documentation,

and the availability of corresponding analysis and extraction

reports. From our perspective, the usability of ERP systems in

this respect might be improved by providing corresponding

standard reports aiding systems administrators. In particular,

this issue pertains to combining all relevant tables for partic-

ular application scenarios and to the matching of events to

underlying process instances. In our case, the latter issue was

exacerbated by the use of differing primary keys in related

database tables. We are not in a position to judge whether

the resulting complexity of the data model is really required.

However, we believe that investigating its actual necessity

might be worthwhile.

With respect to available process mining tools, we found

that there are still certain functions that might be integrated

123

M. Lohrmann, M. Reichert

in such systems to improve their effectiveness. This relates

not only to process improvement projects, but also to other

settings such as compliance management [22] or benchmark-

ing [30]. However, we wish to stress that these issues pertain

to commercially available tools in general. Disco, the tool

used in our project, was chosen as it represents the state of

the art of commercially available tools, in particular regard-

ing ease of use and speed of deployment. Beyond the issues

discussed here, [31] comprises a more detailed summary of

process mining success factors based on multiple case stud-

ies. The following topics should be considered as functions

not yet fully implemented:

–Pattern analysis allows comparing multiple process

enactment variants [32] including their actual frequency.

For example, with regard to repetitive loops (cf. PIM card

5), this functionality would be very useful to identify

and prioritize process variants that should be restricted

or eliminated.

–Compliance rules modeling allows describing relevant

regulations for business processes in a way sufficiently

formalized to automatically check whether these regula-

tions have been adhered to in a process enactment data

sample [22,33]. As an example, consider the requirement

of obtaining approval before job offers are issued.

–Approximation of manual effort facilitates amending

event-based process maps with the underlying manual

processing effort. This would tremendously enhance the

utility of corresponding analyses and could be achieved

by enriching event types with assumptions on the corre-

sponding activities. By matching a material sample log

against total capacity used for processing (the so-called

baselining in practice), the required degree of validity for

the assumptions made could be achieved.

–Automated regression analysis allows finding corre-

lations between characteristics of data samples (e.g.,

between the number of contact partners involved and

cycle times). If characteristics are derived from PIOs,

respective drivers for process improvement can be iden-

tified automatically.

–Sample delineation addresses the issue of restricting a

data sample to exclude process instances without partic-

ular characteristics, such as the presence of start and end

events. Since this topic is important to ensure the valid-

ity of analyses, tool developers might want to consider

guiding users through the sample delineation procedure

by way of an appropriate user interface.

Out of the topics listed above, compliance rules model-

ing and sample delineation can also be addressed through

pattern analysis, which constitutes the basic functionality

to enable process enactment optimization. Like in our sam-

ple case, process improvement projects utilizing empirical

process enactment data will employ spreadsheet applications

if pattern analysis is not readily available in a process mining

tool.

In addition, we used Minitab as an exemplary statistical

tool to support process improvement, e.g., with regard to ana-

lyzing the correlation between cycle times and candidates’

probability to withdraw their applications. This class of tools

can be considered as advanced today and will generally pro-

vide accurate implementations of the relevant statistical tests.

5 Related work

Besides approaches directly addressing the topic, the assess-

ment of PIPs relates to a broad array of fields. These range

from general process improvement and quality to consider-

ations on empirical research on information systems and are

shortly described below.

5.1 Validation of process improvement patterns

Approaches aiming at empirical validation of PIPs can be

traced back to quality management and business process

reengineering approaches which have evolved since the

1950s and the early 1990s, respectively.

In terms of quality management, Six Sigma [27]ispar-

ticularly interesting because it aims at eliminating errors in

manufacturing processes through a set of quantitatively ori-

ented tools used to identify and control “sources of poor

quality.” While the scope of BPM usually lies in adminis-

trative processes instead, there are important analogies since

Six Sigma is based on step-by-step optimization of produc-

tion processes through experimental changes to parameters.

This means that measures are subject to a priori assessment

before they are implemented.

Business process reengineering (BPR), as exemplified in

[6,34], aims at optimizing processes “in the large” instead of

implementing incremental PIPs. Transferring process enact-

ment to an external supplier or customer constitutes a good

example of this paradigm. While the potentials of this dis-

ruptive approach may seem tempting, more recent empirical

evaluation has shown that the risk of projects failing is sub-

stantial [35]. Thus, incremental implementation of individual

PIPs remains a valid approach.

In contemporary BPM [8], proposes a framework to

select and implement redesign practices. As opposed to our

research, this approach does not aim at assessing individual

PIPs for a specific applications scenario, but at efficiently

appraising a broad framework of practices in order to iden-

tify the most relevant propositions. We used earlier results

from the same authors as a source of PIPs to be assessed in

more detail [2].

123

Business process management

The same authors also developed an approach to appraise

BPR practices [36] based on the analytic hierarchy process

(AHP) [37]. This proposition aims at ranking PIPs (or “best

practices”) according to their importance for the organiza-

tion. This enables limiting further considerations to a prior-

itized set of PIPs. In contrast, the goal of our approach is to

assess individual PIPs for a given application scenario based

on a total set of PIPs that should, in the end, be as large as

possible. However, we believe that the approach of step-by-

step refinement of organizational objectives and PIOs might

be used as input to the AHP in terms of scenario-specific

impact criteria.

In addition, [38] proposes PIPs to tackle the findings of

an earlier case study on workflow implementation regarding

issues with team collaboration. While it is not the objective

of this paper to document a general methodology, it can nev-

ertheless be viewed as an approach to prospectively appraise

the implementation of PIPs for a particular application sce-

nario.

5.2 Identification of process improvement patterns

In the following, we shortly summarize the relevant state of

the art with regard to identifying PIPs that may be subject to

assessment.

Besides leveraging PIPs that emerged from the BPR wave

of the late 1990s and early 2000s, there also exist more

recent attempts to provide a basis for process improvement by

appraising perspectives on business process quality based on

the software quality [3,39] or managerial analysis and control

[4]. These approaches result in sets of quality attributes or

characteristics for business processes. Since measures that

aim at fulfilling quality characteristics constitute process

improvement measures, quality characteristics can be viewed

as PIPs as well. In this context, comprehensively validating

the set of quality characteristics provided by an approach

through empirical analysis remains challenging, because it

will be virtually impossible to find practical cases where

the entire set of quality characteristics “adds value.” In this

regard, the present approach instead enables organizations

to validate quality characteristics to be improved specifically

for a particular application scenario.

Benchmarking constitutes an approach widely employed

in practice today [30]. Organizations seek to identify “best

practices” to improve their business processes by comparing

implementation options and results with “peers.” With regard

to specific industries or application fields, the resulting “best

practices” have also been compiled into specialized frame-

works for process management and improvement such as the

IT Infrastructure Library (ITIL) for information management

[40].

In general, contemporary quality management methods

used in manufacturing and logistics (e.g., Poka yoke to elim-

inate potential sources of errors [41]) can provide interest-

ing pointers for improving business processes. A respective

summary of the state of the art in “total quality manage-

ment” (TQM) can be found in [42]. As an example for a

TQM approach, the European Foundation for Quality Man-

agement (EFQM) proposition for achieving “organisational

excellence” can be based on a business process perspec-

tive [43]. However, the underlying evaluation dimensions,

which can be used to identify process improvement poten-

tials, are rather abstract and require general and scenario-

specific interpretation.

Note that research on PIPs addresses the quality of process

models and process implementations in the sense of business

content. In contrast [44–49], exemplify propositions address-

ing process model quality in terms of structure, comprehen-

sibility etc., i.e., the proper representation of actual business

content by model elements.

5.3 Additional aspects

In IS research, there have been diverse propositions to ensure

common standards of scientific rigor in empirical research

such as field experiments or case studies [50,51]. Hevner et

al. [52] summarized empirical “design evaluation methods”

for information systems research. As a basis of this paper,

we chose the requirements summary by Wieringa et al. [24]

due to its concise, checklist-based character, which makes it

readily applicable to research as well as to discussion with

practitioners.

Gregor [53] provided a taxonomy on various types of the-

ory in information systems research. In this context, PIPs

would fall in the category of “design and action” theories

since they give prescriptions on how to construct artifacts (in

this case, business processes). This perspective is interesting

for the purposes of this paper, since it highlights the limita-

tions of treating a PIP as an object of validation, and hence as

a theory, on its own. Rather, whether a PIP is valid as a pre-

scription to implement or change a business process clearly

depends on the relevant application scenario and organiza-

tional context. In line with our propositions, a corresponding

scenario-aware assessment method is required. This method

then constitutes a “design and action” theory.

The top-down approach of deriving PIOs and PIMs from

organizational objectives is similar to techniques for elicit-

ing requirements in goal-oriented requirements engineering

such as KAOS [54]ori*[55]. Propositions in this respect are

based on working with stakeholders to identify goals to be

metbyasystem[56]. Goals are refined step by step until a

level is reached that is suitable for technical implementation.

We stipulated that a structure of PIOs based on organiza-

tional objectives is useful to avoid redundant discussions of

the business value of lower-level PIOs. Similarly, the state of

the art in requirements engineering recognizes the practical

123

M. Lohrmann, M. Reichert

Fig. 12 Approach overview

organizationalorganizational

Understand Understand Refine Identify PIPs

Bundle

relevant PIPs

the application

scenario

organizational

objectives objectives into

PIOs

relevant to

PIOs

into PIMs,

appraise

implementation

benefits of a “goal refinement tree” linking strategic objec-

tives to detailed technical requirements [57]. In this respect,

the concept of organizational objectives compares well to

the notion of “soft goals” [58]. The step-by-step refinement

of PIOs corresponds to the basic AND/OR decomposition

of goals which has been extended to common notations for

goal documentation such as KAOS [54]. These parallels are

based on the shared underlying notion of breaking down a

larger problem, such as overall cost improvement, into more

manageable chunks. This principle can also be found in con-

temporary approaches toward project management, e.g., in

software implementation [59].

6 Discussion

When motivating this paper, we identified three challenges to

be addressed in order to enable a priori PIP assessment. This

section discusses how we addressed these challenges. Fur-

ther, it discusses relevant limitations of our approach and

presents recommendations that may be useful for similar

projects. For quick reference, Fig. 12 recapitulates our propo-

sition as a simplified approach overview: We first seek to gain

a profound understanding of the considered application sce-

nario including the corresponding organizational objectives.

These are then refined into PIOs that are sufficiently granular

to allow identifying relevant PIPs in the next step. Finally,

relevant PIPs are bundled into PIMs and are appraised to

enable implementation decisions.

6.1 Revisiting research challenges

In the previous sections, we described an approach toward

a priori PIP assessment. With respect to Challenge 1 (cf.

Sect. 1.1), we believe that this approach is better suited to

reflect common practice in the field than the available state

of the art in IS research (cf. Sect. 5.1). While there have been

propositions toward ex-post empirical validation of PIPs in

the past, to our knowledge, the approach presented in this

paper constitutes the first proposition toward a priori assess-

ment of PIPs in the area of BPM. In particular, the two per-

spectives on PIP appraisal differ in their treatment of the

available set of PIPs:

–Theex-post perspective seeks to narrow down the set of

PIPs to a selection of aspects with demonstrable empirical

relevance in a wide variety of application scenarios, thus

following common scientific practice.

– The a priori perspective seeks to accommodate a compre-

hensive set of PIPs, but limits assessment to one particu-

lar application scenario. It thus “constructively validates”

only a limited set of PIPs at a time.

Without doubt, the first perspective reflects common sci-

entific practice, since it enables generic statements on PIPs

that are independent of a particular context. Nevertheless, we

found that the second perspective tends to be more in line with

the expectations of practitioners. In our opinion, this reflects

a central characteristic of PIPs and the corresponding impli-

cations for their practical adoption: As becomes clear when

considering PIOs for various application scenarios, charac-

teristics that drive organizational objectives may differ sub-

stantially for varying sample processes. Hence, a validation

of PIPs based on other application scenarios is of limited

value for implementation decisions. In this context, the prac-

titioners we interviewed observed that a preselection of PIPs

will be effective only in the case of frameworks address-

ing a particular field of application. As examples, consider

industry-specific “best practices” such as ITIL for the field

of information management [40], or guidelines for dermato-

oncology in medicine [60].

Assessing PIPs for each individual project requires sub-

stantial effort by qualified personnel to understand the appli-

cation scenario, identify and refine organizational objectives

and PIOs, select appropriate PIPs, and finally bundle them

into implementable PIMs. Whether this effort can be justi-

fied before initiating the assessment depends on the creation

of business value that may be reasonably expected. Organi-

zations should consider three topics before starting a PIPs

assessment project:

– Is the business process substantially relevant to the orga-

nization, e.g., with regard to the output produced or the

cost volume incurred?

– May the organization assume that there are improvement

potentials in the process, for example, when considering

existing problem reports or benchmarks [30]?

– Are there particular circumstances that require analyzing

the process anyway such as, in our case, intentions to

replace the underlying PAIS?

In our sample case, we could assume an overall annual

process cost of about 11.6m Euros (cf. Sect. 4.1). Thus, it

became clear that even minor cost potentials identified would

suffice to cover assessment effort.

123

Business process management

Based on these observations, we believe that our approach

toward PIP assessment is better aligned with common prac-

tice in the field and thus better suited to address Challenge 1

(cf. Sect. 1.1) than the previous state of the art.

Regarding Challenge 2, evaluating our approach through

a substantial experience report, Sects. 2and 4presented the

real-world case we used to this end and the results of apply-

ing our propositions. The sample case dealt with a busi-

ness process found in most organizations and comprised

27,205 cases managed through a standard ERP system. It

exposed typical issues when dealing with empirical analysis

of real-world process enactment data, such as the complex-

ity of extracting and interpreting data, as well as relevant

process characteristics that do not become apparent by ana-

lyzing transactional data. We thus stipulate that our experi-

ence report represents common practical problems fairly well

and has been suitable to evaluate our proposed approach.

To address Challenge 3, the reconciliation of our propo-

sitions to applicable scientific standards, we used a frame-

work by Wieringa et al. [24] to guide the description of our

approach. This enables us to trace all relevant components

of an approach that fulfills scientific criteria, and simplifies

the appraisal whether the corresponding requirements can

be considered as fulfilled. We found that the more rigor-

ous documentation of problem statement and research design

demanded by scientific rigor required some additional effort

in comparison with what is usually found in practice. How-

ever, this task proved still worthwhile, since it simplified final

discussion of proposed PIMs with stakeholders. As an exam-

ple, consider the impact of cycle times on the probability of

candidates to decline job offers, which could only be demon-

strated through rigorous statistical testing.

Overall, we consider our research challenges as met based

on the considerations made above. However, there are still

some relevant limitations we discuss in the following.

6.2 Relevant limitations

We demonstrated the application of our approach on the

basis of a substantial real-world business process with 27,205

process instances. Nevertheless, the first issue that needs to

be discussed with respect to limitations of our approach per-

tains to its evaluation on the basis of only one application

scenario and thus a limited set of relevant PIPs. As a limi-

tation to the environment of data collection (cf. Sect. 3.2),

applicants could not be interviewed because of privacy reg-

ulations. It will thus be useful to apply the approach to addi-

tional scenarios to extend the set of PIPs actually applied.

Of course, this will require access to additional real-world

process improvement projects with substantial sets of empir-

ical data on business processes. To draw meaningful conclu-

sions, these processes should be comparable to what is com-

monly found in other organizations. Accordingly, additional

experience reports (with the potential to extend the under-

lying approach) shall be an issue for future work taking up

such opportunities.

A second topic relates to the availability of a comprehen-

sive set of PIPs to be applied to PIOs. In principle, this does

not affect the validity of our approach. However, it impacts

its practical effectiveness, since it will determine the actual

business value of PIMs identified. In this respect, much work

has been undertaken by Reijers and Limam Mansar [2,8], but

the matter of conceptually deriving a comprehensive set of

PIPs, which might be amended with more concise additions

for specific application fields, still remains an open issue. We

intend to contribute to this topic in the course of our ongoing

work on business process quality [4] by identifying applica-

ble quality attributes for business processes.

On a more abstract level, the third issue pertains to

methodological limitations with respect to empirically val-

idating PIPs. In this context, PIPs can be viewed as a pre-

diction or theory dealing with the impact of certain process

characteristics on process performance. However, compara-

ble to design patterns in software engineering [9], PIPs do

not constitute a self-contained concept for the following two

reasons. As discussed above, currently no approach is avail-

able to demonstrate that a set of PIPs is comprehensive. In

addition, the degree of utility of any given PIP is highly spe-

cific to the application scenario considered. Thus, it is virtu-

ally impossible to validate an entire set of PIPs by means of

empirical information systems research such as field exper-

iments, participative research, or case studies [61]. As dis-

cussed in Sect. 5, researchers have addressed this issue by

conducting meta-studies on a broad range of PIPs [8]. This,

however, means that individual PIPs are validated based on

widely varying research designs. The approach presented in

this paper also cannot resolve this issue. Still, it constitutes

a generally applicable and reusable approach to assess PIPs

for given application scenarios, which can contribute to har-

monize PIP appraisal designs.

A fourth issue that needs to be discussed concerns inherent

limitations with regard to demonstrating the general validity

of the assessment approach proposed. The approach results

in recommendations on which PIPs to implement. However,

the question is how we can ensure that these recommenda-

tions are well founded. This challenge is exacerbated by two

topics:

– On a more detailed level, the business value of PIPs is

appraised considering the business process and the sce-

nario addressed. That is, the general assessment approach

is refined specifically for each application scenario. Thus,

it is not possible to fully replicate the same assessment

approach in other settings, which limits the possibility

of empirical validation. In other words, the validity of

predictions on the business value of a particular PIP in a

123

M. Lohrmann, M. Reichert

particular setting cannot provide assurance on the validity

of predictions on other PIPs in other settings.

– Revisiting PIMs after implementation will only allow

identifying “false positives,” i.e., PIMs that did not

deliver the business value expected. “False negatives,”

i.e., PIPs not chosen for implementation which would

have delivered a positive business value, will always

remain undetected.

Nevertheless, it is still good organizational practice to

track the results of PIM implementation. This provides an

incentive to involved stakeholders to apply due diligence dur-

ing PIPs assessment. However, since only “false positives”

can be tracked, one should be aware that this might lead

to overly risk-averse assessment practices. Setting top-down

process improvement targets (e.g., via quantitative bench-

marking) can be a way to respond to this challenge.

A final limitation takes up the issue of “false negatives”

described above. It pertains to the degree of control we have

with regard to the procedure of selecting PIPs and proposing

PIMs for an application scenario. We have to be aware that

this procedure depends on the knowledge, experience, and

creativity of project participants. In other words, if no project

participant can think of a way in which a PIP could be used

to address a PIO, the PIP will not be considered in PIM

propositions. However, this does not mean that the PIP cannot

provide value in the application scenario. We stipulate that

the step-by-step refinement of PIOs is a useful technique to

address this issue since it helps to focus efforts on relevant

aspects. However, it cannot provide formal assurance on this

issue.

6.3 Recommendations for implementing our method

When working with practitioners to identify and assess PIPs

applicable to our sample scenario, we encountered several

general issues and recommendations that should be consid-

ered when applying PIPs in process improvement projects.

We discussed these observations with our interview partners

in the course of the respective steps in our approach. On

that basis, we phrased a number of project recommendations

that we present in the following. These recommendations

were reconciled with management level interview partners

and may be viewed as guidance for researchers and practi-

tioners when setting up and executing comparable projects.

Readers familiar with these topics may wish to skip this sec-

tion.

Our first recommendation pertains to the overall structure

of the proposed approach and to the “research design” com-

ponent as required in [24].

Project Recommendation 1 (top-down process improve-

ment methodology) Top-down process improvement refers

to methods based on an initial definition of and agreement

on the goals to be pursued, which are then further elabo-

rated and amended with corresponding measures in a step-

by-step approach. As a general principle, earlier decisions

are refined to a more detailed level in later project phases.

Top-down approaches address challenges resulting from two

topics: First, process improvement projects typically require

effective collaboration between multiple parties in an organi-

zation. However, these may tend to “sub-optimize” by focus-

ing on individual interests instead of overall organizational

objectives. As an example, consider the recruiting depart-

ment and the various business units in our sample scenario.

To “sub-optimize” its own effort in application handling, the

recruiting department might pass applications not to the best,

but to the most accessible contact partner in a business unit,

thus impeding the goals of the organization as a whole. To

realize the full potential of process improvement, parties need

to be aligned toward clearly defined common goals and deci-

sions as early as possible. Second, projects without a top-

down decision structure may be obstructed by re-discussing

goals and decisions multiple times. Besides the additional

effort caused, this may lead to inconsistencies in the project.

As an example, consider multiple measures addressing cycle

times. Without a general understanding that cycle times are

an objective of process improvement, this discussion will be

led for each corresponding measure individually. With a top-

down approach, earlier goals and decisions can serve as a

gauge to appraise later decisions and measures.

The top-down principle is reflected in our approach. First,

we require an early senior management agreement on the

general “call for action” (see the concept of organizational

objectives), which is then refined into process-related objec-

tives (see the PIOs concept), and finally into individual

improvement measures. This way, the discussion of individ-

ual measures focuses on how things are to be achieved instead

of what to achieve in general. To implement this recommen-

dation, agreed project results should be strictly documented,

e.g., in a decision log.

The second recommendation is applicable to the “unit”

and “environment of data collection” components as described

in [24].

Project Recommendation 2 (identification of potential

PIOs, PIMs, and PIPs based on process design and enact-

ment) Potentially applicable PIOs, PIMs, and PIPs should

be identified not only by considering the process model, but

also by analyzing empirical data on actual process enactment.

This is crucial to focus on topics of actual value potential.

For example, consider the selection of critical cases in Fig. 3,

which is reflected in the PIOs for our sample case.

The third and fourth recommendations address data gath-

ering and analysis procedures required to appraise PIOs and

PIMs for a particular application scenario. In terms of [24],

they qualify “measurement” and “data analysis procedures”.

123

Business process management

Project Recommendation 3 (appropriate qualitative or

quantitative demonstration of business value) For each

PIO to be addressed by PIMs, the underlying business value

must be empirically demonstrated based on proper qualita-

tive or quantitative analyses with respect to organizational

objectives. Likewise, the business value of PIMs must be

made transparent through appropriate analyses.

The specific analytic approach for individual PIOs and

PIMs must consider the actual application scenario, balanc-

ing expected insights against analysis efforts. For example,

the omission of tasks that obviously do not contribute to the

business objective of the process can be justified by a short

qualitative description. In contrast, the introduction of addi-

tional control tasks to diminish defects later on in the process

will require careful quantitative weighing of pros and cons.

Project Recommendation 4 (identifying relevant stake-

holders as interview partners) To ensure the validity of

measurement procedures, proper selection of interview part-

ners is particularly relevant for PIOs and PIMs that should be

validated qualitatively. For BPM scenarios, it is important to

interview experienced senior personnel overlooking the end-

to-end business process and to represent both the “supplier”

and the “customer” perspective to avoid lopsided optimiza-

tion. For our sample process, we interviewed the head of

recruiting operations and the administrator of the applica-

tion management process from the “supplier” side, and the

HR partner of a business unit as well as team managers from

the business unit from the “customer” side.

The fifth and sixth recommendations concern the final

assessment of PIMs. Hence, they refer to “data analysis pro-

cedures” as well [24].

Project Recommendation 5 (considering implementation

effort in business value appraisal) When discussing the

business value of particular PIMs for a business process, the

respective implementation effort must be taken into account.

This includes measures required, cost, time, and change man-

agement issues (e.g., training personnel to enact new activi-

ties). A PIM will only provide business value if implemen-

tation effort is justified by realized process improvement

potentials. For example, an organization may demand that

the required investment must not exceed three times the pro-

jected annual cost savings when appraising operational cost

optimization measures.

Projectrecommendation6(leveraging“quickwin”poten-

tials) In many practical scenarios, it is possible to identify

“quick win” PIMs that can be implemented with limited effort

and should thus be given higher priority than others, in par-

ticular in comparison with full-scale PAIS implementation

measures which are usually very costly. Examples include

the elimination of process defects caused by process partic-

ipants’ behavior, interface issues between departments, or

issues of data quality. Note that these topics are often identi-

fied through empirical analyses (e.g., using process mining).

7 Summary and outlook

This paper described an approach for a priori, scenario-

specific assessment of process improvement patterns based

on organizational objectives, process improvement objec-

tives, and process improvement measures. In our approach,

we leveraged available work on generic requirements toward

empirical research in IS engineering [24]. We thus demon-

strated how these principles can be applied to practical cases

while ensuring the general appeal of our approach.

We reported on the application of the approach to a real-

world business process, including validation of the respective

results with practitioners. The approach led to the identifica-

tion of fivepotential process improvement measures that bun-

dle and refine individual process improvement patterns for

the given application scenario. Matching the expected gains

against implementation and operating efforts, the organiza-

tion was enabled to take well-informed implementation deci-

sions. Revisiting the proposed process improvement mea-

sures more than 1year after initial data collection confirmed

that these decisions could be used to guide further develop-

ment of the business process in practice.

In future research, we will integrate our PIP assessment

approach into a broader proposal to manage the quality of

business processes [4]. Moreover, we will apply it to further

real-world application scenarios to gain additional experi-

ence. This will enable us to further validate and elaborate the

approach. The assessment of PIPs as well as the validation

of PIOs and PIMs will always require to refine our general

approach according to the application scenario. However,

considering additional sample scenarios might enable us to

define standard types of assessment and validation proce-

dures, for example, based on corresponding types of PIOs

that occur in multiple application scenarios. In turn, this may

further facilitate the practical adoption of our approach.

OpenAccess This article is distributed under the terms of the Creative

Commons Attribution License which permits any use, distribution, and

reproduction in any medium, provided the original author(s) and the

source are credited.

References

1. Mutschler, B., Reichert, M., Bumiller, J.: Unleashing the effective-

ness of process-oriented information systems: problem analysis,

critical success factors and implications. IEEE Trans. Syst. Man

Cybern. (Part C) 38(3), 280–291 (2008)

2. Reijers, H.A., Liman Mansar, S.: Best practices in business process

redesign: an overview and qualitative evaluation of successful

redesign heuristics. Omega 33(4), 283–306 (2005)

3. Heravizadeh, M., Mendling, J., Rosemann, M.: Dimensions of

business processes quality (QoBP). In: Proceedings of the Busi-

ness Process Management Workshops (BPM Workshops 2008).

LNBIP, vol. 17. Springer, pp. 80–91 (2009)

123

M. Lohrmann, M. Reichert

4. Lohrmann, M., Reichert, M.: Understanding business process qual-

ity. Number 444 in studies in computational intelligence. In: Busi-

ness Process Management: Theory and Applications. Springer, pp.

41–73 (2013)

5. van der Aalst, W.M.: Reengineering knock-out processes. Decis.

Support Syst. 30, 451–468 (2001)

6. Davenport, T.J., Short, J.E.: The new industrial engineering: infor-

mation technology and business process redesign. Sloan Manag.

Rev. 4, 11–27 (1990)

7. Zur Muehlen, M., Ho, D.T.: Service process innovation: a case

study of BPMN in practice. In: Proceedings of the 41st Hawaii

International Conference on System Sciences (HICSS’08), IEEE

Computer Society, p. 372 (2008)

8. Limam Mansar, S., Reijers, H.A.: Best practices in business process

redesign: use and impact. Bus. Process Manag. J. 13(2), 193–213

(2007)

9. Gamma, E., Helm, R., Johnson, R., Vlissides, J.: Design pat-

terns: abstraction and reuse of object-oriented design. In: Proceed-

ings of the 7th European Conf on Object-Oriented Programming

(ECOOP’93). LNCS, vol. 707. Springer, pp. 406–431 (1993)

10. IDS Scheer: Process intelligence white paper: What is process intel-

ligence? (2009), http://www.process-intelligence.com. Accessed 7

Apr 2010

11. van der Aalst, W.M.P.: Process Mining: Discovery, Conformance

and Enhancement of Business Processes. Springer, Berlin (2011)

12. Reichert, M., Weber, B.: Enabling Flexibility in Process-

Aware Information Systems: Challenges, Methods, Technologies.

Springer, Berlin (2012)

13. Lohrmann, M., Reichert, M.: Demonstrating the effectiveness of

process improvement patterns. In: Proceedings of the 14th Working

Conf on Business Process Modeling, Development, and Support

(BPMDS’13). LNBIP, vol. 147. Springer, pp. 230–245 (2013)

14. Lohrmann, M., Reichert, M.: Modeling business objectives for

business process management. In: Proceedings of the 4th S-BPM

ONE—Scientific Research. LNBIP, vol. 104. Springer, pp. 106–

126 (2012)

15. The Object Management Group: Business Process Model and

Notation: Version 2.0 (January 2011), http://www.omg.org/spec/

BPMN/2.0. Accessed 12 Oct 2011

16. Günther, C.W., Rozinat, A.: Disco: discover your processes.

In: Processing of the Demonstration Track of the 10th Interna-

tional Conference on Business Process Management (BPM’12).

Springer, pp. 40–44 (2012)

17. Grigori, D., Casati, F., Castellanos, M., Dayal, U., Sayal, M., Shan,

M.C.: Business process intelligence. Comput. Ind. 53(3), 321–343

(2004)

18. van Dongen, B.F., Alves de Medeiros, A.K., Verbeek, H.M.W.,

Weijters, A.J.M.M., van der Aalst, W.M.P.: The prom framework:

a new era in process mining tool support. In: Proceedings of

the 26th International Conference on Application and Theory of

Petri Nets (ICATPN’05). LNCS, vol. 3536. Springer, pp. 444–454

(2005)

19. Celonis GmbH: Celonis Discovery: Overview, http://www.celonis.

de/en/softwarelosungen/celonis-discovery/uberblick/. Accessed

19 Sept 2013

20. Simon, H.A.: The Sciences of the Artificial, 3rd edn. MIT Press,

Cambridge (1996)

21. March, S.T., Smith, G.F.: Design and natural science research

on information technology. Decis. Support Syst. 15(4), 251–266

(1995)

22. Ly, L.T., Knuplesch, D., Rinderle-Ma, S., Goeser, K., Pfeifer, H.,

Reichert, M., Dadam, P.: SeaFlows toolset—compliance verifica-

tion made easy for process-aware information systems. In: Pro-

ceedings of the CAiSE’10 Forum—Information Systems Evolu-

tion. LNCS, vol. 6412. Springer, pp. 332–346 (2010)

23. Ittner, C.D., Larcker, D.F.: Assessing empirical research in manage-

rial accounting: a value based management perspective. J. Account.

Econ. 32(1–3), 349–410 (2001)

24. Wieringa, R., Heerkens, H., Regnell, B.: How to write and read a

scientific evaluation paper. In: Proceedings of the 17th IEEE Inter-

national Requirements Engineering Conference (RE’09), IEEE

Computer Society, pp. 361–364 (2009)

25. Society for Human Resource Management: Executive brief:

What factors influence cost-per-hire? (2012), http://www.shrm.

org/benchmarks. Accessed 3 Nov 2012

26. Lohrmann, M., Riedel, A.: Process quality and performance in

shared services organizations. In: Finance Bundling and Finance

Transformation. Springer Gabler, pp. 225–251 (2013)

27. Pyzdek, T., Keller, P.: The Six Sigma Handbook. McGraw-Hill,

New York (2009)

28. Adam, E.E., Corbett, L.M., Flores, B.E., Harrison, N.J., Lee, T.,

Rho, B.H., Ribera, J., Samson, D., Westbrook, R.: An international

study of quality improvement approach and firm performance. Int.

J. Oper. Prod. Manag. 17(9), 842–873 (1997)

29. Lohrmann, M., Reichert, M.: Efficacy-aware business process

modeling. In: Proceedings of the 20th International Conference on

Cooperative Information Systems (CoopIS’12). LNCS, vol. 7565.

Springer, pp. 38–55 (2012)

30. Camp, R.C.: Benchmarking: The Search for Industry Best Prac-

tices that Lead to Superior Performance. Quality Press, Milwaukee

(1989)

31. Mans, R., Reijers, H., Berends, H., Bandara, W., Prince, R.: Busi-

ness process mining success. In: Proceedings of the 21st Euro-

pean Conference on Information Systems (ECIS’13). Paper 102

(2013)

32. Li, C., Reichert, M., Wombacher, A.: Mining business process vari-

ants: Challenges, scenarios, algorithms. Data Knowl. Eng. 70(5),

409–434 (2011)

33. Knuplesch, D., Reichert, M., Ly, L.T., Kumar, A., Rinderle-Ma,

S.: Visual modeling of business process compliance rules with the

support of multiple perspectives. In: Proceedings of the 32nd Inter-

national Conference on Conceptual Modeling (ER’13). LNCS, vol.

8217. Springer, pp. 106–120 (2013)

34. Hammer, M., Champy, J.: Reengineering the Corporation. A Man-

ifesto for Business Revolution. HarperBusiness, New York (1993)

35. Cao, G., Clarke, S., Lehaney, B.: A critique of BPR from a holistic

perspective. Bus. Process Manag. J. 7(4), 332–339 (2001)

36. Limam Mansar, S., Reijers, H.A., Ounnar, F.: Development of a

decision-making strategy to improve the efficiency of BPR. Expert

Syst. Appl. 36(2), 3248–3262 (2009)

37. Saaty, T.L.: How to make a decision: the analytic hierarchy process.

Eur. J. Oper. Manag. 48(1), 9–26 (1990)

38. Reijers, H., Poelmans, S.: Re-configuring workflow management

systems to facilitate a smooth flow of work. Int. J. Coop. Inf. Syst.

16(2), 155–175 (2007)

39. Heinrich, R., Paech, B.: Defining the quality of business processes.

In: Proceedings of the Modellierung 2010. LNI P-161, Bonner

Köllen, pp. 133–148 (2010)

40. Best Management Practice: IT Infrastructure Library: 2011 Edition

(2011), http://www.itil-officialsite.com/. Accessed 17 Apr 2012

41. Fisher, M.: Process improvement by poka-yoke. Work Study 48(7),

264–266 (1999)

42. Dale, B.G., van der Wiele, T., van Iwaarden, J. (eds.): Managing

Quality, 5th edn. Wiley-Blackwell, Oxford (2007)

43. European Foundation for Quality Management: An Overview

of the EFQM Excellence Model (2012), http://www.efqm.org/

en/PdfResources/Overview%20EFQM%202013%20v1.pdf.

Accessed 21 Dec 2012

44. Becker, J., Rosemann, M., von Uthmann, C.: Guidelines of

Business Process Modeling. LNCS, vol. 1806. In: Business

123

Business process management

Process Management: Models, Techniques, and Empirical Stud-

ies. Springer, pp. 241–262 (2000)

45. Vanderfeesten, I., Reijers, H., van der Aalst, W.M.: Evaluating

workflow process designs using cohesion and coupling metrics.

Comput. Ind. 59(5), 420–437 (2008)

46. Mendling, J., Reijers, H.A., van der Aalst, W.M.P.: Seven process

modeling guidelines (7PMG). Inf. Softw. Technol. 52(2), 127–136

(2010)

47. Weber, B., Reichert, M., Mendling, J., Reijers, H.: Refactoring

large process model repositories. Comput. Ind. 62(5), 467–486

(2011)

48. Reijers, H., Mendling, J.: A study into the factors that influence the

understandability of business process models. IEEE Trans. Syst.

Man Cybern. Part A 41(3), 449–461 (2011)

49. Reijers, H., Mendling, J., Dijkman, R.: Human and automatic mod-

ularizations of process models to enhance their comprehension. Inf.

Syst. 35(5), 881–897 (2011)

50. Benbasat, I., Zmud, R.W.: Empirical research in information sys-

tems: the practice of relevance. MIS Q. 23(1), 3–16 (1999)

51. Jedlitschka, A., Ciolkowski, M., Pfahl, D.: Reporting Experiments

in Software Engineering. In: Advanced Topics in Empirical Soft-

ware Engineering. Springer (2007)

52. Hevner, A.R., March, S.T., Park, J., Ram, S.: Design science in

information systems research. MIS Q. 28(1), 75–105 (2004)

53. Gregor, S.: The nature of theory in information systems. MIS Q.

30(3), 611–642 (2006)

54. Dardenne, A., van Lamsweerde, A., Fickas, S.: Goal-directed

requirements acquisition. Sci. Comput. Program. 20(1–2), 3–50

(1993)

55. Yu, E.S.K., Mylopoulos, J.: Using goals, rules, and methods to

support reasoning in business process reengineering. Int. J. Intell.

Syst. Account. Finance Manag. 5(1), 1–13 (1996)

56. Nuseibeh, B., Easterbrook, S.: Requirements engineering: A

roadmap. In: Proceedings of the Conference on The Future of Soft-

ware Engineering (ICSE’00), pp. 35–46 (2000)

57. van Lamsweerde, A.: Goal oriented requirements engineering: a

guided tour. In: Proceedings of the 5th IEEE International Sympo-

sium on Requirements Engineering (RE’01), IEEE, pp. 249–262

(2001)

58. Soffer, P., Wand, Y.: On the notion of soft-goals in business process

modeling. Bus. Process Manag. J. 11(6), 663–679 (2005)

59. Balzert, H.: Lehrbuch der Softwaretechnik: Basiskonzepte und

Requirements Engineering. 3rd edn. Spektrum Akademischer Ver-

lag in German (2009)

60. van der Geer, S., Reijers, H., Krekels, G.: How to run an effective

and efficient dermato-oncology unit. J. German Soc. Dermatol.

8(1), 15–18 (2010)

61. Jenkins, A.M.: Research Methodologies and MIS Research. In:

Research Methodologies and MIS Research. Elsevier, Amsterdam,

pp. 103–117 (1985)

Matthias Lohrmann holds a

diploma degree in business infor-

mation systems from Regens-

burg University, having grad-

uated in 2002. He works as

a consultant for general and

administrative processes such as

finance, HR, and IT. In particu-

lar, Matthias advises large cor-

porate clients in the manufac-

turing industry. Complementary

to his professional occupation,

Matthias has been working on his

PhD thesis on quality aspects of

business processes since 2009.

Manfred Reichert is professor

at the University of Ulm, Ger-

many and director of the Data-

bases and Information Systems

Institute. His major research

interests include business process

management (e.g., adaptive pro-

cesses, data-centric processes,

knowledge-intensive processes,

and mobile processes), service-

oriented computing (e.g., service

composition), and process-aware

information systems in advanced

application domains (e.g., auto-

motive engineering). Manfred

pioneered the work on the ADEPT process management technology

and is co-founder of the AristaFlow GmbH. He has been participat-

ing in numerous research projects in the BPM area contributing more

than 250 scientific papers on BPM-related topics. His book entitled

“Enabling Flexibility in Process-Aware Information Systems”, which

he co-authored with Barbara Weber, was recently published by Springer.

Manfred has been PC Co-Chair of the BPM’08, CoopIS’11, and

EDOC’13 conferences and General Chair of the BPM’09 and EDOC’14

conferences.

123