This Working Paper Series is published by the Green
Consumption Assistant (GCA) Research Project.
The Green Consumption Assistant supports consumers in making more sustainable decisions
during online shopping. The GCA displays green product alternatives on the search engine Ecosia
and provides information about more sustainable alternatives, for example, references to repair,
rental, or sharing options. In addition, sustainable websites will be highlighted on Ecosia and the
climate commitments of the organisations and companies will be made transparent in a ranking.
For the recommendations of the GCA, a comprehensive product database (Green Database) with
ecological and social sustainability information is being built up using machine learning
techniques.
The GCA is a collaboration project between the Technische Universität Berlin, the Berliner
Hochschule für Technik, and the green search engine Ecosia and is funded by the
Bundesumweltministerium as a lighthouse project for artificial intelligence in use for ecological
challenges. The project embodies a new, interdisciplinary partnership that combines sustainable
and behavioral research with machine learning, user-centered design, and digital product
development.
In the project, we rely on cooperation and exchange with various sustainability actors, scientists,
and label organisations or online shops, to ensure a reliable and comprehensive data set for the
recommendations of the Green Consumption Assistant.
All issues of the GCA Working Papers Series can be downloaded free of charge at: https://green-
consumption-assistant.de/en/publications/#working-paper
The GCA Working Paper Series serves to publish initial results from the ongoing research project
and is intended to promote the exchange of ideas and academic discourse.
More information:
Website: green-consumption-assistant.de
Contact: info@green-consumption-assistant.de
Citation: Maike Gossen, Jens Bergener, Patricia Jankowski [2023]: #07 Sustainability Tips,
Eco-Quiz or AI-powered Chat? Exploring the perception of sustainable behavior change
interventions on Ecosia using focus groups. GCA Working Paper Series, Berlin.
https://doi.org/10.14279/depositonce-20020
DOI: https://doi.org/10.14279/depositonce-20020
This research was funded by the German Federal Ministry for the Environment, Nature
Conservation, Nuclear Safety and Consumer Protection, grant number 67KI2022A.
License: This work is licensed under a Creative Commons Attribution 4.0 International
License. More information, see https://creativecommons.org/licenses/by/4.0/
Sustainability Tips, Eco-Quiz or AI-powered Chat?
Exploring the perception of sustainable behavior change interventions on Ecosia
using focus groups.
Working Paper for the GCA Research Project by researchers of the TU Berlin
December 2023
Maike Gossen, Jens Bergener, Patricia Jankowski
Abstract
User research shows that Ecosia users desire guidance on sustainable lifestyles when searching
for information on the green search engine. However, it is important to better understand the
sustainability information needs of different user groups to design compelling environmentally-
related search engine features. Therefore, we conduct a qualitative study investigating how
users perceive different behavior change interventions on Ecosia’s search engine. We use three
interventions (sustainability tips, eco-quiz and AI-powered chat) and discuss prototypes with six
online focus groups, each consisting of people from different sustainability-oriented
backgrounds (‘environmental awareness types’). After performing content and sentiment
analysis, the results show that the perception of the features as attractive or unattractive is
contingent on the presentation of the feature as well as individual factors, such as environmental
awareness. The sustainability tips have received a positive response from Ecosia users as well
as Open-minded, Sceptics and Undecided user groups. The AI-powered chat evokes polarised
discussions and is perceived as attractive by Open-minded, Consequentialist and Undecided
users. Conversely, only Open-minded and Ecosia users found the eco-quiz attractive. The
qualitative study underlines the importance of user segmentation and targeting to improve the
effectiveness of behavior change interventions. The study suggests improving the tested
prototypes on Ecosia and offers insights into the sustainability information needs of more
sceptical and undecided user groups.
Keywords
Behavior change interventions, sustainable behavior, sustainability features, search engine,
qualitative research, focus groups
TABLE OF CONTENTS
Introduction ............................................................................................................................ 1
Method ................................................................................................................................... 2
Design ................................................................................................................................. 2
Material ............................................................................................................................... 3
Participants ......................................................................................................................... 5
Fielding ............................................................................................................................... 7
Analysis ............................................................................................................................... 7
Qualitative content analysis .............................................................................................. 7
Sentiment Analysis........................................................................................................... 8
Findings .................................................................................................................................. 8
The role of online search engines in daily life (RQ1) ................................................................ 8
The use of search engines as a resource for sustainability-related information (RQ2) ........... 10
The perception of sustainability features on Ecosia (RQ3) ................................................... 12
Sustainability Tips .......................................................................................................... 12
Eco-Quiz ........................................................................................................................ 15
Ecosia Chat .................................................................................................................... 18
Discussion and conclusion ..................................................................................................... 22
General discussion of main findings ................................................................................... 22
Discussion of findings with regard to the different groups .................................................. 24
Methodological limitations ................................................................................................. 27
Practical implications ........................................................................................................ 27
References ........................................................................................................................... 29
1
INTRODUCTION
Most people feel that more needs to be done to combat climate change. Generally, research
reveals that a large group of respondents is on the cusp of more environmentally oriented
behavior (Federal Ministry of the Environment and German Environmental Agency, 2023).
However, the same study also shows the uncertainty and demands of people in clarifying what an
effective personal contribution to climate protection can look like and how it should be made
(Federal Ministry of the Environment and German Environmental Agency, 2023). At the same
time, most people continue to say they are already doing everything they can to protect the
environment, suggesting many consumers feel they have reached a limit in their perceived ability
to help make an impact (GlobeScan, 2023). These findings are consistent with previous research
in the Green Consumption Assistant project. User research and usability tests at Ecosia have
shown that users of the green search engine express the desire to receive instructions for a more
sustainable life when searching for information online. They also want more environmentally-
related search engine functions that stand out from other conventional search engines and
underline their demand for personalised sustainability information.
After an ideation phase with Ecosia's designers, engineers and the project team's sustainability
experts, several features were designed and developed as part of the Green Consumption
Assistant. The design phase was informed and influenced by insights from behavioral change
research such as the Behavior Change Wheel (BCW), which provides a valuable and systematic
framework for behavioral change interventions (Michie et al., 2011). The BCW is based on
psychological guidance for behavior change interventions in various thematic fields, including
environmental sustainability. It consists of three layers: the hub around which nine interventions
and seven policy categories are placed that could enable the interventions to occur. The hub of
the wheel provides a simple framework for understanding sustainable behavior, including three
essential conditions: capability (e.g., knowledge and skills), opportunity (e.g., time and cultural
norms) and motivation (e.g., habits and analytical decision-making) which must be satisfied in
order for a desired behavior to occur (Michie et al., 2011). Capability and motivation both refer to
internal factors specific to the individual, whereas opportunity refers to the external
environment. In the model, these sources of behavior are surrounded by nine interventions that
address deficits in one or more of the central conditions. The interventions include education,
persuasion, incentivisation, coercion, training, restriction, environmental restructuring,
modelling, and enablement. For example, the psychological capability for sustainable behavior is
linked to education, training and enablement. In addition to the conceptual contribution of the
BCW, the design of sustainability features was particularly inspired by the Behavior Change
Technique (BCT) Taxonomy, including 93 techniques in 16 behavior change categories (Michie et
al., 2013). Of the various techniques, the techniques in the ‘shaping knowledge’, ‘reward’ and
‘comparison of behavior’ categories were the most informative and helpful for the design of the
sustainability features. For example, by providing guidance on performing a certain behavior
through sustainability tips (‘shaping knowledge’), users are supported in acquiring knowledge and
expertise about a certain sustainable behavior. When the tips are combined with a poll, users’
attention is drawn to the performance of others to enable comparison with their behavior or to
aspire to or imitate the behavior of others (‘comparison of behavior’). The ‘social reward’
technique inspired us to design an eco-quiz that rewards correct answers verbally. Lastly, we
developed an AI-driven conversational agent concept, Ecosia Chat, to provide climate and
2
environmental information related to common user questions, integrating different behavior
change techniques depending on the user's information request.
To examine the perception of Ecosia’s sustainability features, we conducted six focus groups
with individuals of different sustainability-oriented backgrounds. Five groups represent a
specific environmental awareness type based on the Environmental Awareness Study 2020 and
show characteristic patterns concerning the sub-domains of these types – the Sceptics, the
Undecided, the Open-minded, the Oriented, and the Consequentialists (Stieß et al., 2022). In
addition, a group of Ecosia users was formed based on the criteria of Ecosia use and combined
screening criteria for a mix of environmental awareness types. Focus groups can be useful when
exploring “perceptions, feelings, and thoughts about issues, ideas, products, services, or
opportunities” (Krueger & Casey, 2014, p. 37). Group discussions aim to seek ‘everyday
explanations’ during group interactions (Calder, 1977) and to learn more about attitudes and
opinions in an informal group setting (Hyden & Bulow, 2003). Within sustainability research, focus
groups are suitable for evaluating and further developing ecological products, services and
communication strategies to align product design as closely as possible with user requirements
and identify needs for information and action (Henseling et al., 2006). In addition to investigating
perceptions of Ecosia’s sustainability features, we also wanted to discuss opinions on search
engines as a general source of information and sustainability information in particular. The
following research questions guided the qualitative study:
RQ 1: How do participants perceive the role of online search engines in their daily lives?
RQ 2: How do participants use search engines as a resource for sustainability-related information?
RQ 3: How do participants with different sustainability-oriented backgrounds perceive
sustainability features of Ecosia?
The remainder of this working paper is organised as follows. In the next section, we provide
information on the focus groups, followed by the presentations of findings. The final section
integrates and discusses the findings, derives practical implications, and outlines the study's
limitations and avenues for future research.
METHOD
Design
The online focus groups were conducted using synchronous audio and video-based methods,
which have become increasingly popular in the social sciences. The COVID-19 pandemic has only
accelerated this trend, leading to greater use of virtual formats and web conferencing platforms.
To ensure the success of our online focus groups, we followed the recommendations and
technological considerations of Pocock et al. (2021). For example, we developed privacy and
confidentiality standards and conducted practice sessions to ensure that all facilitators were
comfortable using the audio-visual platform. Additionally, we obtained informed consent from
participants and prioritised data security. We used a discussion guide with a selected group of
questions and discussion points to structure the focus group sessions and to encourage
meaningful discussion. The questions were developed and refined through feedback from
experts in the field and tested during a pilot session. During the group discussions, participants
were encouraged to engage in open and dynamic conversation with their fellow group members
and the moderator. The moderator utilised a direct and active approach to exploring the topics
3
at hand, with each group starting with an introductory icebreaker question before transitioning
to more focused questions related to the study's goals. Following the introductory section,
stimulus material was employed to elicit reactions from participants. Our study focuses on the
perception of the presented prototypes, primarily at the attitudinal level of spontaneous
opinions, intuitions, and beliefs. Across all groups, a consistent moderating style was
maintained.
Material
During our study, we used prototypes of the sustainability features created by Ecosia. As
mentioned in the Introduction, we tested two versions of daily sustainability tips (one with a poll)
and an eco-quiz. Both features are displayed when users open a new tab in their browser and
have selected Ecosia as their default search engine (see Figure 1 and 2). While the sustainability
tips are a dynamic feed of sustainability-related content, the eco-quiz asks questions about
sustainability. Both concepts are designed to encourage individuals to think and reflect on
sustainable behavior in their everyday lives. They convey direct and indirect problem knowledge
about a specific sustainable behavior and provide instructions on carrying it out. The third
feature, the ChatGPT-powered conversational agent Ecosia Chat (see Figure 3), enhances the
user’s search experience within the main Ecosia search engine vertical. Ecosia Chat is designed
to provide climate and environmental information in the context of common user questions. The
functionality and design of the features are illustrated in the following figures.
Figure 1: Screenshot of the demo video for sustainability tips on Ecosia’s start page opening a
new tab using dark mode (here including the poll).
4
Figure 2: Screenshot from the demo video for the eco-quiz on Ecosia’s starting page after
opening a new tab using dark mode.
5
Figure 3: Screenshots from the demo video for the Ecosia Chat vertical feature and exemplary
sections of three stages of conversation with the generative AI presented during the focus
groups.
The prototypes of the features were the first iteration of a step-by-step development process
and ready for implementation and field testing on Ecosia. In the focus groups, each prototype
was demonstrated with a demo video via screen sharing while the moderator explained the
functionality of each feature.
Participants
Participants were recruited through a panel institute and compensated for their participation.
To prevent potential biases, they were invited to participate in a survey on "Internet and everyday
life" instead of focusing solely on environmental topics. We aimed to explore ‘collective
orientations’ (Bohnsack, 1989). Therefore, we opted for purposeful sampling and homogeneous
composition for each focus group to ensure respondents shared similarities regarding their
social and economic backgrounds. According to Warr (2005, p. 201-202), this is beneficial
because “participants who belong to pre-existing social groups can recreate aspects of their
social relations as participants exchange opinions and engage in the process of persuasion”. A
screening questionnaire was used to determine eligibility.
1
The selection process was based on
predefined socio-demographic characteristics, attitudes towards the environment and climate
change, environmental behaviors, and willingness to change. In total, we conducted six online
focus groups with six participants each. Five groups represent a specific environmental
awareness type based on the Environmental Awareness Study 2020 and show characteristic
patterns concerning the sub-domains of these types – the Sceptics, the Undecided, the Open-
minded, the Oriented, and the Consequentialists (the Rejecters were excluded in our study for
practical reasons, Stieß et al., 2022). In addition, a group of Ecosia users was formed based on
the criteria of prior Ecosia use and combined screening criteria for a mix of environmental
awareness types. The people who participated in the Ecosia focus group had ticked Ecosia when
asked which search engine they use or have used in the past during the screening process.
However, during the discussion it emerged that they are more likely to use Ecosia as their
secondary search engine, while still using Google as their primary. Therefore, they are referred
1
The screening questionnaire can be provided on request.
6
to as new or casual Ecosia users. Table 1 shows the socio-demographic characteristics, attitudes
towards the environment, climate change awareness, environmental behaviors, and willingness
to change for each of the focus groups. According to the findings of the 2020 Environmental
Awareness Study (Stieß et al., 2022), four of the six focus groups (Open-minded, Oriented,
Consistent and the mixed type of Ecosia users) have a high level of awareness of the environment
and climate change. They also have at least a medium level of environmental behavior and are
willing to change their behavior for the good of the environment. According to the study, these
groups prioritise environmental and climate protection and are open to a socio-ecological
transformation. The study also suggests that these groups have a high level of approval for more
restrictive measures and interventions. On the other hand, the Undecided group shows a certain
openness towards climate issues, but is expected to be rather reluctant to change their behavior.
Finally, the Sceptics group, on the other hand, is expected to adopt a negative or distanced
attitude towards transformation and behavioral measures.
Table 1: Socio-demographic characteristics, attitudes towards the environment, climate change
awareness, environmental behaviors, and willingness to change for each of the focus groups.
Subdimensions of environmental awareness
Socio-demographic characteristics
No.
Group
name
Environment
al attitude
Climate
awareness
Environmen
tal behavior
Willingness
to change
Gender
Age
Education
Income
Place of
residence
1
(casual)
Ecosia
users
High to very
high
High to very
high
Medium to
high
Medium to
high
2 male, 4
female
23 - 34
2 medium,
4 high
1 low, 2
middle, 2
high, 1 very
high
1 small town,
2 middle to
small town, 3
big city
2
Sceptics
High to very
high
Medium
Medium
Medium
4 male, 2
female
39 - 51
Medium
High
2
countryside,
2 small town,
2 middle to
small town
3
Undecided
High to very
high
High
Low to
medium
Medium
4 male, 2
female
21 - 56
Medium
1 low, 5
medium
2
countryside,
4 middle to
small town
4
Open-
minded
Very high
Very high
Medium
High
2 male, 4
female
23 - 35
High
2 low, 2
high, 2 very
high
1
countryside,
3 middle to
small town, 1
big city
5
Oriented
Very high
High
Medium to
high
Medium to
high
2 male, 4
female
31 - 58
1 medium,
5 high
3 high, 3
very high
2
countryside,
4 big city
6
Consequen
tialists
Very high
Very high
High
High
2 male, 4
female
32 - 50
1 medium,
5 high
4 high, 2
very high
2 Middle to
small town, 4
big city
7
In addition, all participants were screened to ensure that they had experience with the daily use
of search engines. The findings cannot be generalised for the German population as the sample
(N=36) is relatively small. However, the homogeneous group composition enables a cross-group
comparison.
Fielding
The focus groups took place in August, 2023, via a web-conferencing software. Each group
lasted 90 minutes and was moderated by a trained project team member who was assisted by a
co-moderator
2
. All focus groups were audio and video recorded and protocolled. The protocols
focussed on the literal content of what was said and ignored certain expressions of the spoken
word. The researchers involved in the study also took detailed discussion notes that were
incorporated into the analysis to contextualise the narrative of the transcripts. The accuracy of
qualitative data depends on the circumstances under which it was collected (Schulz et al., 2012).
For example, focus group participants may intentionally or unintentionally distort their
experiences and opinions. Therefore, a thorough analysis of the focus group discussions is
necessary to uncover possible biases. In order to achieve optimal validity, we subjected the
discussion guide to a qualitative pre-test in which we applied the think-aloud method. In addition,
external team members were tasked with reviewing the protocols of the focus groups for
distractions or possible bias.
Analysis
Qualitative content analysis
The focus group discussions were transcribed and then analysed using MaxQDA software
(version 2022). The data analysis followed qualitative content analysis standards inspired by
Mayring’s (2020) approach. We established a category system in accordance with the research
objectives. To extract key themes and categories within the data, we used both inductive and
deductive coding methods (Strauss & Corbin, 1998). The primary themes of the analysis were
formed by the pre-defined discussion protocol using deductive coding, while inductive coding
extracted the nuances within these themes. We created a coding guide and systematically
grouped the codes together. Before finalising the coding guide, two researchers independently
reviewed a randomly selected focus group for coding consistency to ensure reliability. Two
researchers coded each focus group separately, then compared their codings and resolved any
differences. After the protocols were processed using the category system, the material was
paraphrased, reduced, and summarised in several steps. The detailed process of coding and
summarising is illustrated in the flow model in Figure 4. This approach allowed for an in-depth
analysis and interpretation of the material by facilitating cross-comparison of all subgroups.
2
The moderation guide can be provided on request.
8
Figure 4: Flow model
Sentiment Analysis
In order to record and quantify the attitudes and opinions of the participants on the prototypes,
we analysed the data from the focus groups using MaxQDA's sentiment analysis. MaxQDA
classifies the focus group data based on sentiment polarity using a lexicon that contains a
sentiment score for each word in the lexicon. Selected focus group segments are then
categorised and automatically coded as positive, slightly positive, neutral, slightly negative, or
negative sentiment. The sentiment labels were reviewed and changed if needed. This way, we
investigate which affective associative spaces the prototypical features evoke. Affective
associations in the negative range indicate that a feature is perceived as unattractive, positive
values correspond to favourable perception, and neutral values signify indifference. Additionally,
using this method, we were able to identify and assess the emotions, feelings and opinions of
different focus groups on a particular characteristic or topic.
FINDINGS
The role of online search engines in daily life (RQ1)
Search engines are perceived by all participants as reliable companions and central sources of
information. They are seen as uncomplicated, fast and helpful. Users start their search query
with more or less specific goals: It is often initially used to find ideas (e.g. for gifts, recipes) or to
check current conditions (e.g. weather, opening times), followed by further research and
comparisons towards concrete planning. The search topics cover all aspects of daily life, from
everyday questions and advice, personal interests and leisure activities to services and product
comparisons. Although using a search engine is seen as time-saving, several participants
mention that they often get lost in a search and spend significantly more time than planned.
9
Participants in all groups use search engines habitually and take a multi-step, iterative approach
to their search, starting with a general search and using the information obtained to narrow down
the results further. Very often, search is only used to navigate to desired websites. It is believed
that search engines determine the order of results somewhat arbitrarily. These are therefore
often questioned and individual criteria are created to classify websites as trustworthy. It is
noticeable that the usage practices of the Open-minded, Oriented and Consequentialists groups
appear to be more advanced (this could be related to the higher educational status in these
groups): Search queries are very targeted and systematic. Participants are aware of the impact
of search terms on the quality of search results and use appropriately savvy search terms.
Foreign languages and other search engine functions such as image search or other search
engines are used to supplement the results.
Which device is used for the use of search engines depends in all groups on availability and
current needs. The smartphone is generally used for short, targeted searches on the go, while
laptops or desktop computers are clearly preferred for longer searches, online shopping or travel
planning due to the larger screen size. The search engine, which is usually identical for both
mobile and desktop use, is considered the first point of contact when searching for information.
Well-known and trustworthy websites are only selected directly for individual, personally
relevant topics such as hobbies or news. Selected apps on smartphones replace the search for
information via a search engine. This applies in particular to pre-installed applications (e.g.
weather, routes) and/or selected, frequently used services (e.g. banking, emails, social media).
When asked about the ideal search engine, all groups agreed on some key criteria: the relevance
and accuracy of search results is considered to be by far the most important criterion. The idea
of "first time right", i.e. fast and accurate results on the first search, is of great importance.
Google is considered the best search engine in this respect. Google is also frequently used for
searches among Ecosia users (including for sustainability topics). This shows that the relevance
of the results takes precedence over other factors. Efficiency and time saving are also important
criteria. Participants appreciate the simplicity of the search, including features such as auto-
complete and operators that help refine search queries. The summary generated above the
search results is also appreciated. Despite security concerns and knowledge of the collection of
data by search engines, most participants do not consider data protection to be a decisive
(exclusion) criterion. The situation is similar with advertising and sponsored content. Although
these are perceived as annoying, there is a conflict regarding the willingness to pay for search
engine services, which is why advertising is accepted as a necessary evil. Desirable
improvements compared to previous search engines are the topicality of search results and the
associated filtering out of outdated websites and information. Aspects of personalization or a
unique selling point of the search engine are occasionally mentioned.
The criteria already mentioned play a decisive role in the choice of search engine. Across all
groups, Google is the dominant search engine due to its speed, relevance of search results and
personalised algorithms that presort results. It is also perceived as less ad-heavy than
alternatives and is the default search engine pre-installed on many devices. Nevertheless, there
are some concerns about data protection and the need to register. Bing, on the other hand, is
mainly used out of convenience when it is pre-installed on purchased devices. In some cases, it
is also deliberately used as an alternative to Google, especially when users want to receive less
personalised information. Bing's results are considered less trustworthy and reputable than
Google's, but the search engine has become more interesting for some users with the
10
introduction of AI Chat. Other search engines are rarely used. Yahoo, for example, is usually only
used occasionally out of habit or because of integrated services such as email accounts.
Alternative search engines such as DuckDuckGo or Good are only used as long as the relevant
core topic (e.g. data protection, resource consumption) remains in people's minds. The tested
search engine Ecosia is only known in a few groups. The concept of using part of the revenue for
climate protection projects (e.g. planting trees) is rated positively across the board (“I think it's
good that trees are planted there when I search for something. It gives you a good feeling that you
are doing something good.”
3
, Sceptics), but the actual use is low. This is often justified by the
actual or assumed lower relevance of the results compared to conventional search engines (“I
don't think it's a bad idea, but I'd be worried if it was as information-rich as Google, that would be my
requirement, it's probably difficult.”, Oriented).
Ecosia's search results are perceived by former users as inferior in quality and quantity (less
relevant content) compared to Google, while non-users express similar concerns without being
seen. However, participants express a (slight) interest in Ecosia due to its German company
status and better compliance with EU data protection law.
It is interesting to note that ChatGPT is increasingly being used as an alternative to search
engines, especially for complex questions. Among other things, participants appreciate the
ability to refer back to previous conversations and searches. However, using ChatGPT requires
an understanding of how language models work, such as correct input and communication, which
not all participants are familiar with.
The use of search engines as a resource for sustainability-related information
(RQ2)
Search engines play an important role in the search for information on sustainability. When
search engines are visited for topics related to sustainability, it is usually to get a rough
orientation and overview. Search engines are primarily used for more complex sustainability
topics and specific information needs, e.g. for information research in the run-up to significant
purchases (“I am a bit further along in the sustainability topic and know what I am looking for, even
if I am looking for something completely different.”, Consequentialist). The search results are
checked as far as possible for their seriousness and comprehensibility and only the links that
appear to fulfil these criteria are clicked on. In some cases, participants had developed their own
strategies for checking information and sources. The trustworthiness and credibility of
information plays a major role in the use of search engines and is addressed and discussed by the
participants in all groups.
Without direct search interest, information on sustainability is not visible in search engines, with
a few exceptions (such as Google's theme days). Instead, sustainability topics are more likely to
be known from social media (”I don’t search for sustainability topics on search engines at all, I tend
to follow people on Instagram and pick up something from them.” Ecosia users) or from other
websites and the participants encounter them (analogously) in stores. In one group (Ecosia
users), doubts about the credibility of search results were cited as a reason for using information
channels other than search engines. Occasionally it is discussed (in the groups Undecided, Open-
minded and Oriented) in connection with sustainability that the use of search engines itself
3
All direct quotes in this working paper are translated from German to English by the authors.
11
requires energy and causes CO2 emissions. The search engine providers are held responsible for
dealing with this responsibly.
Figure 5: Diagram for the frequency distribution and number of sentiments towards
sustainability and search engines across all focus groups.
Overall, the sentiment analysis suggests a positive relationship between participants' opinions,
sustainability information within search engines and their affective evaluation (see Figure 5). In
the discussions about specifically displayed sustainability information in search engines,
certain applications were rated attractive in all groups as long as they contained valuable
information, did not convey moralising or patronising messages and did not restrict user-
friendliness. A distinction is made between topics and search queries, as not all content is
suitable for being accompanied by additional sustainability information. The participants discuss
banners and pop-ups as applications for which user-friendliness is essential. Only the group of
Consequentialists does not expect corresponding sustainability applications from a search
engine as they do not appear conceivable to them on their search. In addition (or instead), labels,
icons, or filters for sustainability information and a counter for search queries (and the
associated energy consumption) were discussed as attractive suggestions (“[...] would set a
filter, I only search for sustainable things or things that are socially correct. I want to determine that
myself and Google will tell me”, Consequentialists). In some cases (Oriented and
Consequentialists), the credibility and traceability of the sources of the sustainability
information displayed are emphasised. For participants, this transparency is important to
navigate and avoid misinformation or greenwashing.
12
The perception of sustainability features on Ecosia (RQ3)
Sustainability Tips
The general assessment and evaluation of the sustainability tips (with or without poll) are
relatively positive across all groups. At the same time, there are neutral and negative sentiments
towards the feature (see Figure 6). Participants are positive about the opportunity to receive
ideas, food for thought and inspiration on sustainability topics on a daily basis without having to
look for them specifically. These opportunities can promote an unconscious awareness of
sustainability issues. Regarding the potential to trigger changes in behavior, all groups rated the
tips quite suitable, as they present uncomplicated and low-threshold consumption alternatives
with high everyday relevance.
Figure 6: Diagram for the frequency distribution and number of sentiments for the Sustainability
Tips feature across all focus groups.
The separate assessment and evaluation of the sustainability tips uncovers differences between
the groups (see Figure 7). Overall, there are more positive statements for the sustainability tips
in the groups of (casual) Ecosia users, Sceptics and Open-minded. While the number of
sentiments are evenly distributed between positive and negative statements, in the group of the
Undecided, they tend to be slightly positive in their overall sentiment. In contrast, the
sustainability tips are perceived slightly negatively in the groups of the Oriented and
Consequentialists.
13
Figure 7: Diagrams for the frequency distribution and number of sentiments for the Sustainability Tips feature for each of focus groups.
14
The positive comments on the sustainability tips in the groups cover several topics. The
Undecided particularly prefer new and innovative topics to provide an impetus for changes in
behavior (“I think that's good, because we can not go on like this forever”, Undecided). They also
particularly like the design of the feature. The group of Ecosia users confirms that the
sustainability tips are compatible with the Ecosia search engine if they do not disrupt the user
experience, are placed subtly or discreetly and can be used/ignored on their own initiative. They
also appreciate the opportunity to provide feedback and the learning nature of a broad (and non-
individualized) selection of topics. The reliability of the sources and content is particularly
important to them for the credibility of the tips. The tips match the participants' impression of
Ecosia and are suitable for interacting with the users and binding them to the search engine. The
Open-minded consider the tips particularly effective for users who are already motivated to
change their behavior and are looking for easy access to information and inspiration. On the other
hand, the novelty value and usefulness of the daily tips may be low for these people in particular.
The Oriented find that the sustainability tips are a nice and fun idea as a daily reminder of
sustainability. The tips remind them that even small actions can make a difference if enough
people participate. At the same time, they wonder whether Ecosia users are the right target
group or whether the tips should not be aimed more at people less interested in sustainability.
Especially the Oriented and Consequentialists do not want to be distracted from their actual
search and fear feeling patronised when using the feature (“It's a search engine, not a know-it-all
machine, and [I] do not want to be told off”, Oriented). We found similar statements from some
Ecosia users and participants in the group of Open-minded. The possibility of the tips disrupting
the user experience (similar to pop-ups and banner ads) and interrupting or distracting from the
actual search is perceived as negative by almost all groups. The Oriented also find the tips
“annoying” and “excessive”, for them, the tips are nothing “earth-shattering”. In contrast to the
Open-minded, they also emphasise that people who are already interested in sustainability derive
less benefit from it. The Consequentialists criticise the “raised index finger” and the “time-
consuming” use of the feature. For some of the Sceptics in particular, the seemingly “inflationary”
use of sustainability content reduces the novelty value and the “wow effect” of the feature. In
addition, some Sceptics fundamentally question the purpose of the tips and consider the content
to be “superfluous”.
Almost all groups perceive the addition of a poll to the sustainability tips as disruptive, boring or
unnecessary. The Undecided fear that they will be demotivated by seeing survey results if they
are not yet living very sustainably (“When I saw the first video, I was euphoric, it would have
triggered a point, curiosity. Now I feel trapped [...]. That dampens my euphoria.”, Undecided). Both
the Sceptics and the Consequentialists rate the results as quickly boring because the format
offers them no added value. Only the Open-minded discuss the benefits of the poll in length and
a more differentiated way. For example, the fact that social comparison with others is motivating
for them and can encourage them to change their behavior (“Especially if a lot of people do it If you
think about it, it's not hard to implement, maybe I will do it after all.”, Open-minded). On the other
hand, the poll results can also create a sense of social pressure and frustration among the Open-
minded if they perform poorly in the survey. For this group, it also depends on the topic and its
perceived relevance whether they ultimately take part in the poll. There is also an increased
concern for privacy in this group.
The assessments of the sustainability tips within the groups reflect the general interest in using
the feature, especially in the groups that have a slightly more positive sentiment towards the
15
feature (Ecosia users, Sceptics, Undecided, Open-minded). The Undecideds, for example,
recognize the feature's potential for behavior change if the sustainability information provided is
relevant, interesting, innovative, surprising and easy to implement (“[...] I think the daily tip is good
even without interaction. If I feel like it, I can do it, if not, I do not. It gives me an impulse, an impetus,
I can implement it, I do not have to do anything except read a little”, Undecided). Interestingly, even
the Sceptics expressed an overall positive use intention. Overall and in comparison to the other
features, the sustainability tip performs best in terms of user interest.
The suggestions for improvement of the sustainability tip feature mainly relate to its placement
and the content of the sustainability tips. Participants also suggest additional functions for the
feature. Regarding placement, some participants point out that they rarely visit the start page of
their search engine. Instead, they often search directly via their browser bar. Therefore, they
recommend placing the tips elsewhere or making them findable via a toggle or button on the
search results page. Almost all groups suggested the individualisation and personalisation of the
tips (e.g. by creating a user profile) to display content relevant to their interests (“[...] maybe
something user-defined will be suggested to me.”, Consequentialists). Participants view a reward
system as a playful and motivating element in which users receive points for implementing the
tips. In line with the tip, location-based offerings and information can also be offered via links (in
some cases, participants expressed concerns regarding the data privacy of a location/place
sharing function). Some participants advocate for an on/off button that allows them to control
the display of the tips and turn the feature on or off themselves. Repeating popular tips after
some time, such as through a regular recap or review of the most popular tips is viewed as a
helpful guideline. Additionally, a searchable archive function, that allows users to review tips
themselves, would improve their accessibility. Overall, the text per tip should be short, as
participants state the effort to read the text is too great. The content should focus on everyday,
easy-to-implement actions and the monetary savings potential of sustainable actions. In
addition, possible implementation hurdles have to be taken into account and communicated
proactively.
Eco-Quiz
The overall assessment and evaluation of the eco-quiz is slightly negative across all groups (see
Figure 8). Generally, there are more negative than favourable comments.
16
Figure 8: Diagram for the frequency distribution and number of sentiments for the eco-quiz
across all focus groups.
Comparing the reactions of the individual groups, there are significant differences (see Figure 9).
The tenor of the discussions on the eco-quiz is predominantly negative for the Sceptics, Oriented
and Undecided. At the same time, Open-minded and (casual) Ecosia users weigh up the positive
aspects, while Consequentialists express a balanced amount of positive and negative signals
towards the feature.
17
Figure 9: Diagrams for the frequency distribution and number of sentiments for the eco-quiz feature for each of focus groups.
18
The positive statements across the groups cover several topics. For Ecosia users, the focus is
on dealing with sustainability topics in a playful and entertaining way, as well as on acquiring
knowledge and raising awareness of the importance of sustainable behavior. Similar to the
sustainability tips, the everyday relevance is important for the participants so that the eco-quiz
can trigger changes in behavior. While the Undecided have slightly more negative sentiments
towards the feature, they acknowledge the learning effect and its possibility to raise awareness
for sustainability issues. In a group comparison, the Open-minded were the most positive. They
emphasised the added value of acquiring knowledge, partly thanks to the additional
explanations. Evaluating and classifying their own quiz results can be motivating because they
learn something new. For the Consequentialists, the eco-quiz is an entertaining activity with a
learning effect.
Overall, however, the negative comments and concerns predominate. The Sceptis, Undecided,
Oriented and Consequentialists primarily rated the risk of distraction and the lack of added value
to their search journey as an unfavourable feature of the quiz (“Time is money, it's great to do a
quiz, I don't see the added value and it's not suitable.”, Oriented). For a quiz, according to the
Undecided, better alternatives are available on the market. Additionally, for them the information
is only remembered for a short time because the focus is on the playful element. For the
Consequentialists, the quiz has too little substance and relevance to motivate users to visit the
Ecosia search engine daily. Other negative assessments refer to the lack of incentive (i.e. in the
form of a prize, Sceptics) or the short, superficial and too simple explanations (Ecosia users). For
the Oriented, the quiz is not appealing as it is designed for young people who appreciate and are
used to playful elements.
Accordingly, the interest in using the eco-quiz is relatively low in all groups. Most participants do
not care and would not pay attention to the eco-quiz. Even in the group of Open-minded, who are
most likely to express a positive opinion, the use of interest depends heavily on the context. Only
a few appreciate the opportunity to acquire new knowledge playfully.
The suggestions for improvement relate mainly to the placement and presentation on the
website and additional functions. Firstly, keeping the text shorter may improve readability.
Second, providing the option to click away or move the quiz may allow users to complete it later.
Other ideas include rewards for taking the quiz, such as planting a tree for each correct answer.
Lastly, participants suggest the option of personalising the content to increase its relevance.
Ecosia Chat
To elicit spontaneous associations with Ecosia Chat, participants were asked for their initial
thoughts about a Chatbot on the Ecosia search engine before the actual demonstration of Ecosia
Chat (i.e. just a screenshot of the Chat windows is shown without explaining its functionality). In
general, the initial interpretation of Ecosia Chat’s purpose tends to vary based on participants'
existing knowledge. Some participants mistakenly associate the Chat with Chat and service
assistants often used on websites of banks. Other participants assume it may be used as a
support tool to facilitate searches and evaluate search queries. In some groups, a connection to
sustainability is established (e.g., the Oriented, who expect it to answer certain questions in a
dialog with details and sustainability information), in other groups the participants do not initially
expect this.
After a demonstration of the functionality of Ecosia Chat, opinions vary from enthusiastic to
neutral and negative. The general assessment and evaluation of Ecosia Chat is relatively evenly
19
distributed between positive and negative associations across all groups, while the number of
positive sentiments slightly outweighs the number of negative sentiments (see Figure 10).
Figure 10: Diagram for the frequency distribution and number of sentiments for Ecosia Chat
across all focus groups.
When comparing the reactions of the individual groups towards the Ai-powered Chat, distinct
differences emerge (see Figure 11). (casual) Ecosia users overall express slightly negative
sentiments about an Ai-powered Chat feature. This negativity is even more pronounced within
the group of Oriented users. At the same time, the Open-minded acknowledge a slightly positive
attitude towards the feature, while the Undecided and Consequentialists groups are actively
considering its positive aspects. Meanwhile, the Sceptics group provides a balanced
assessment, acknowledging both positive and negative facets.
20
Figure 11: Diagrams for the frequency distribution and number of sentiments for Ecosia Chat for each of focus groups.
21
Within the Undecided group, the similarity to ChatGPT, is considered positively. For them,
getting to use a similar application to ChatGPT, that is integrated into a search engine without
having to register or create a user profile, makes Ecosia Chat an attractive alternative that also
enables immediate use. The Undecided group highlighted the interactivity and the human-like
conversation style. Additionally, they (alongside the Consequentialists) positively emphasised
the convenience of swiftly acquiring information or details simply by asking questions within
Ecosia Chat. The Consequentialists rate the condensed information within the responses (e.g.,
product comparisons, travel) as particularly helpful and time-saving, as it eliminates the tedious
search for the right information in conventional search engines for them (“It saves time and I think
that's a good thing. [...] It's much more targeted and tailored.”, Consequentialists). The Open-
minded group also expresses a predominant positive opinion towards the feature, although they
are slightly less enthusiastic when compared to the Undecided and Consequentialists. For Open-
minded users Ecosia Chat adds value to their search and allows for a wide range of uses that
make everyday life easier. Particularly, the ease of access to an AI Chat without registration and
the possibility to store Chat histories (when compared to normal search queries) is seen as an
advantage. However, Chat responses have to be of high quality and come from reputable
sources. Within the Open-minded group, the Chat is seen as particularly beneficial when
inquiring about recipes, as it allows to go beyond merely providing or finding a recipe. For
example, it allows them to obtain additional useful information such as a shopping list within the
same conversation. While the Oriented highlight more negative aspects of the chat, particularly
tech-savvy participants from this group, who already have experience with ChatGPT, describe
the feature as a visionary concept (“It's a cool vision that people can find exactly what they are
looking for via the Chat. [...] I imagine the future going in this direction.”, Oriented). Interestingly,
the less favourable groups of Ecosia users and Sceptics, would also consider the sustainability
approach of Ecosia Chat positively if responses would remain unbiased and balanced.
Paternalism due to the sustainability focus of Ecosia Chat is the main concern and source of
negative sentiments within almost all groups (as with the other two features). The group of
Ecosia users discussed that it is intrusive and moralising to suggest information on sustainability
without being asked. Similar concerns are expressed within the Sceptics group. The strong
(emotional) rejection of the sustainability focus might stem from selecting an example from the
food sector, a subject known for its emotional resonance, during the demonstration of the
feature. The topic seemed to be very emotionally charged within the groups (”I like to eat meat,
so when I get a vegan recipe, I immediately have a guilty conscience. [...] I would like to have the
option of eating a vegetarian diet, but I think it's really crazy to just throw away a vegan dish like
that.”, Sceptics). Among the Sceptics, it is evident that participants without prior experience with
ChatGPT find the added value of Ecosia Chat compared to a “normal” search query to be
somewhat unclear. While the Undecided share a general mistrust of applications based on
artificial intelligence, members of these groups are still open-minded and express interest in
trying out such applications. The risk of paternalism is not discussed within this group. For the
Open-minded, the lack of versatility in the Chat’s responses is considered a drawback when
compared to normal search queries. In addition, for them a conversation with Ecosia Chat is
expected to take longer because fewer options are presented at the same time and there are no
images to illustrate the results. In the scenario of the recipe example, there's a risk that the
suggested content may not align with the user's preferences, diminishing its relevance. This can
ultimately limit the perceived quality and relevance of the responses of the Chat. Additionally,
within this group, people without an interest in sustainability expressed the possibility that they
22
could feel patronised or criticised due to the sustainability focus of the responses (although not
quite as clearly negative as in other groups). At the same time, the references to sustainability
within the responses are occasionally perceived as too commonplace to provide substantial
added value. The Oriented group, who are generally interested in sustainability topics, find that
using the chat requires a basic understanding of how to formulate a prompt. For them, the default
focus on sustainability topics is too strong and intrusive (“[...] but I am not even asked what do you
actually want, something is thrown out right away.”, Oriented). Instead, this group suggests that
users should be able to ask questions about sustainability based on their interests. Similarly, the
Consequentialists express uncertainty about the added value of Ecosia Chat compared to a
simple search engine search. They also criticise the persistent reference to sustainability, which,
for them, limits the relevance of the responses, with some participants finding this aspect
annoying. (”The fact that the conversation doesn’t come with tips, but rather with hints, think about
it again, gives me food for thought, not a raised index finger, which is what I would expect.”,
Consequentialists).
Following their generally negative assessment of the feature, the groups of Ecosia users,
Sceptics and Oriented expressed little or no interest in using Ecosia Chat. Conversely,
particularly the tech-savvy participants within the Undecided, Open-minded and
Consequentialists groups express a comparatively high level of interest in testing and using the
feature. Participants within these groups who are less tech-savvy also exhibit a general
openness towards using the feature. However, their enthusiasm is somewhat more subdued.
The feedback collected for improvement of Ecosia Chat is primarily based on negatively
perceived aspects of the feature. All groups emphasised that the relevance of the responses to
their prompts should be the top priority and that sustainability topics should be integrated more
passively in the background, as food for thought. The sustainability references within the
responses of the Chat could be suggested as an alternative, for example through a query or pre-
selection process with different relevant topics, rather than being included in all responses as a
default. In the same way, the personalization of responses is perceived as useful, for example,
when responses and sustainability suggestion are provided based on user's interests,
preferences, or Chat history, making it more helpful and relevant to them (“[...] if he asks how you
like to eat, vegan or do you have an allergy, then you can choose one.”, Sceptics). As far as the
tonality of the Chat responses is concerned, participants suggest avoiding a moralistic approach.
In addition, participants rate a context-dependent display of direct links and images as useful.
Other useful improvements include a read-aloud feature, the ability to save or store responses
and send them via email. Additionally, participants ask for clear instructions on how to properly
use the AI Chat, such as prompting examples. Another improvement for participants would be to
make the selection of responses more transparent to users, so that they may understand the
underlying mechanism of response generation.
DISCUSSION AND CONCLUSION
General discussion of main findings
The focus groups have shown that search engines generally play an important role in the search
for information in everyday life. This confirms what is known from prior research. For example,
online users primarily rely on search engines and online marketplaces to discover e-commerce
sites for online purchases (The Future Shopper Report, 2022). Search engines, in particular,
23
provide users with a convenient and rapid overview of product choices as they explore and
evaluate shopping information. Furthermore, many users trust search engines (Lewandowski,
2012). Search engines are increasingly facing competition from language models and
conversational agents such as ChatGPT. This development was already evident in the focus
groups, although the enthusiasm for these new technologies and the interest in using them was
mainly expressed by tech-savvy participants.
According to the focus groups, with a few exceptions, sustainability information can hardly be
found in search engines if users do not search for it directly. Instead, participants are more likely
to come across sustainability topics on social media and from influencers they already follow. As
information on sustainability is currently rarely found in search engines, features that provide
specific information on sustainability topics are generally perceived as useful and attractive
when searching for general information to show ways to adopt a more sustainable lifestyle. The
positive evaluation of the topic by the groups, even in the groups that were expected to be more
reluctant to change their behavior (Undecided) and have a negative or distanced attitude towards
transformation and behavioral measures (Sceptics), suggests that the focus of the discussion
should be less on whether search engines should display sustainability information and more on
how, what, where (at what point in the search journey) and for whom sustainability interventions
should be designed.
Against this background, the perception of the presented feature prototypes is mixed, with
rather positive feedback on the sustainability tips, balanced and nuanced reactions to the Ecosia
Chat and rather negative evaluations of the eco-quiz. In the following, we summarise the
assessment of the sustainability features. Almost all groups insist on prioritising functional over
environmental benefits when it comes to the sustainability of the presented features on a search
engine. For example, they prioritise satisfying their core information needs over environmentally
related actions by pointing out the discrepancy between the immediate and everyday problems
they are trying to solve through a particular search and the somewhat abstract environmental
responsibility in these scenarios. When confronted with the sustainability features, participants
do not initially think about their impact on future environmental behavior. Instead, they often
focus on the impact on their daily routines and the disruption to their usual search engine
information routines. Interestingly, the features trigger a similar defensive reaction in all focus
groups. The defence mainly refers to what could be understood as behavioral impositions (Mau
et al., 2023). Above all, respondents resist the feeling that they are being told how to live their
lives when they conduct a normal, everyday search. These concerns shed light on an implicit
expectation users have of search engines: Autonomy and self-determination. As a result,
participants feel threatened and disrupted in their everyday search practices and respond by
defending their habits and preferences, their consumption choices and their data privacy against
the influence of others. Interestingly, it is not necessarily the content of the interventions (the
what) that is at the centre of concern at this point, but the perceived obligation, coercion,
interference and paternalism (the how) of the prototypes in all focus groups. This view is
expressed by respondents from different social backgrounds. In social psychology, the
corresponding impulse is referred to as reactance (Rosenberg & Siegel, 2018), i.e. an impulsive
rejection of demands on one's own lifestyle that are perceived as moral or ecological. The above
findings suggest questioning the social constraints and tone of sustainability communication
approaches that primarily emphasise reflexivity and target people's everyday lives. While
sustainability information is generally perceived as useful and attractive in showing ways to a
24
more sustainable lifestyle, a reaction to certain sustainability features can be expected from
those who fear perceived behavioral impositions, but also from those who insist on their
behavioral autonomy, self-determination, and the validity of their common sense on
environmental issues (see Mau et al., 2023).
Discussion of findings with regard to the different groups
By segmenting the focus groups based on their environmental awareness types, this study was
able to determine which of the groups can be targeted with the presented features, and for which
groups these approaches may not be effective. In the following, we summarise the evaluation of
the features presented for each group separately. This also includes a brief description of the
group characteristics based on the work of Stieß et al. (2022).
The environmental awareness type of the Sceptics is characterised by an attitude towards the
environment that is slightly below average, but in absolute terms it is at a high level. Their climate
attitude is only moderate and their environmental behavior is at an average level. The willingness
to change, on the other hand, is relatively weak. They are expected to have a negative or distant
attitude towards socio-ecological transformation and behavioral measures. Therefore, this
group is among those who are probably the most difficult to reach through environmental
communication. Against this background, the ambivalent feedback on the features from the
participants in this focus group comes at no surprise. While the eco-quiz is perceived negatively
and not much helpful, the Ai-powered Chat elicits a polarised discussion with both positive and
negative opinions. However, overall, daily sustainability tips are perceived as slightly positive
within the group. In general, it is noticeable that the discussions are very emotional and
participants’ comments (e.g., “superfluous”, ”useless”, “inflationary”) indicate that they attach
little importance to the features presented and do not see them as having a big influence on their
own behavior. Thus, this group is likely to remain difficult to reach with sustainability
interventions on search engines. However, easy-to-understand and personally relevant
sustainability messages (e.g., in the form of daily sustainability tips) seem to be the most
promising and suitable intervention for this group as long as they are non-restrictive and backed
up with evidence. At the same time, for users who have already had experience with chatbots, an
AI-powered chat can be a promising tool for answering the Sceptic’s complex questions about
sustainability, as long as the answers are unbiased and balanced.
The environmental awareness type of the Undecided can be described by a fairly high
environmental attitude as well as a high climate attitude. Their environmental behavior, on the
other hand, is rather weak. This means that there is a particularly strong discrepancy between
attitude and behavior in this group. The Undecided are particularly important due to its size in the
population and their positive attitude towards environmental and climate protection. They
recognize the urgency of environmental problems, but place more emphasis on other challenges
such as social justice. Environmentally and climate-friendly behavior seems to be most affected
by obstacles in this group compared to the others, e.g. lack of knowledge for the implementation
of climate-friendly behavior. In general, Undecided users value the presentation of sustainability
information within a search engine as long as it does not compromise user-friendliness. The
feedback for all three features is balanced. Undecided users are not explicitly negative about any
of the features and, due to their share of the population, represent an important target group for
Ecosia and sustainability interventions in general. Overall, they show a slightly positive tendency
towards the AI-powered Chat and sustainability tips and a slightly negative tendency for the eco-
25
quiz. For them, sustainable options that are available where they normally search are particularly
helpful. On the other hand, the interactivity, the human-like conversational style and the
convenience of quickly obtaining information within the chat function meet their information
needs in terms of sustainability. The AI-driven chat thus breaks down some of the barriers to
more environmentally and climate-friendly behavior in this group, e.g. by addressing their lack of
sustainability knowledge in a conversational style. In summary, implementing measures that take
into account the socio-demographic circumstances of individuals who are indecisive about
adopting environmentally friendly behaviors and providing them with relevant knowledge about
how to implement these behaviors in a non-restrictive way can help increase their acceptance of
sustainability interventions.
The environmental awareness type of the Open-minded is characterised by a high environmental
and climate attitude while their environmental behavior is moderate. A large gap between
attitude and behavior can be observed also in this group. Compared to most other groups, they
believe that not only politicians and citizens, but also businesses, have a greater responsibility to
do something to protect the climate. They are also predominantly of the opinion that climate
change cannot be stopped with technology alone and without a change in lifestyle. They should
therefore be particularly open-minded and interested in behavior change interventions such as
the sustainability features of Ecosia. In general, they consider the display of sustainability
information in a search engine to be useful if it contains relevant and non-patronizing
information. Accordingly, the sustainability features presented are predominantly met with high
approval in this group. Within this female-dominated group, we can observe an ecological ethos
of voluntary sustainable action that is typical of the young and educated middle class, who “want
to do” and “can do” more for the environment. This makes the Open-Minded an ideal target group
for sustainable features on a search engine. The participants appreciate the user-friendliness
and practical benefits of all the presented prototypes, which, on the other hand, leads to a more
critical perception of the features if they would not provide relevant information or take up too
much time (as is feared in the case of Ecosia Chat).
The environmental awareness type of the Oriented is characterised by a very high to high
environmental awareness and high environmental behavior. Moreover, they are only on average
willing to make (further) changes. In view of their already comparatively strong environmental
behavior, one reason for this could be that they have already implemented some changes, but
need support in order to be able to implement further (possibly more difficult) changes. The
group of the Oriented therefore represents an important target group for low-threshold
behavioral change interventions. In general, Oriented users find features with sustainability
information useful if they meet their specific expectations, if they promise credibility and
traceability, and if they avoid misinformation and greenwashing. Interestingly, this focus group
tends to evaluate all the sustainability features presented rather negatively. While they think the
features are a fun idea to remind themselves of sustainability and their prior knowledge, they do
not want to be distracted from their actual search and fear that they will feel patronised when
using them. Furthermore, the playful eco-quiz does not offer sufficient added value for their daily
search. In line with their strongly oriented views and environmental awareness, the Oriented
group finds the predetermined sustainability aspect in the answers of an AI-driven chat intrusive
and the sustainability information presented often irrelevant for people with their prior
knowledge. At the same time, they perceive the chat as a future-proof application and value its
current focus on sustainability as a relevant application for people who lack prior knowledge of
26
sustainability topics. Given their comparatively strong environmental knowledge and their
indications that the features in their current form seem more suitable for users with less
sophisticated knowledge, future iterations of the presented interventions should communicate
their specific capabilities and advanced features to provide support for implementing further
(possibly more difficult) changes and to address the needs of more oriented user groups.
The last type of environmental awareness, the Consequentialists, have the strongest
environmental awareness compared to all other groups. They are strongly oriented towards
environmental protection and sustainable behavior. People in this group are sensitive to climate
protection and open to even more far-reaching behavioral changes in favour of sustainability. In
general, Consequentialists do not expect corresponding sustainability applications from a
search engine. However, they perceive sustainability information features as useful, if they cover
their specific and strongly oriented information needs related to sustainability. In line with their
general opinion about sustainability information on search engines, their assessment of the
presented sustainability features is mixed, with rather critical feedback on the sustainability tips
and the eco-quiz, but enthusiastic feedback for the Ai-powered Chat in its current form. For
them, features such as the sustainability tips are somewhat insufficient, insubstantial and
boring, precisely because their own sustainable consumption behavior is already quite advanced.
In contrast, the Ecosia Chat is perceived as visionary and extremely practical because it provides
prompt and helpful answers to their sustainability information needs. In contrast to the Oriented,
they perceive the sustainability requirements in the answers as less restrictive. Nevertheless,
the Consequentialists express concerns that the chatbot's focus on sustainability could limit the
quality of its answers. They also wondered whether the chatbot would remain neutral when
answering questions that do not concern the environment. In general, the AI-powered chatbot
primarily meets their need to search for and identify information on sustainability themselves
without the search engine having to intervene. Additionally, for them, the condensed information
within the chat responses is helpful as it eliminates the conventional search process within the
search engine and saves time.
In the discussions, we found that although the group of Ecosia users has experience with Ecosia,
most of the participants in this group do not use Ecosia as their default search engine. Most of
the participants in this group have tried the search engine but prefer other search engines such
as Google because they were not always satisfied with the quality of the search results of their
queries on Ecosia. Thus, we contextualize this group as new or casual Ecosia users. Therefore,
the results can not be generalised for regular or power users of the search engine. The groups’
environmental awareness is a mixture of most other environmental awareness types with high to
very high environmental attitudes and climate awareness and medium to high environmental
behavior and willingness to change. In addition, this group is comparatively young, female-
dominated and the participants have a high standard of education. Generally, this group tends to
rely on sources other than search engines when it comes to sustainability information, such as
social media, websites or information in stores at the point of sale. The reason for this is their
perceived uncertainty about the credibility of search results and the efficiency of the search
process. If sustainability interventions such as banners, pop-ups, icons/symbols and labels are
informative and credible, do not detract from the user experience and are not judgmental, then
they are generally attractive to this group. In line with their general attitude towards
sustainability information in search engines, they perceive sustainability tips and the eco quiz as
slightly positive. For them, the tips and the playful eco-quiz fit well with the Ecosia brand and are
27
perceived as helpful as long as they are relevant to everyday topics. On the other hand, these
Ecosia users perceive an AI-powered Chat rather negatively (negative comments also dominate
the group comparison). Their main concern is the integration of sustainability references without
being asked or having the opportunity to make a choice, as they expect Ecosia’s chat to have
similar features to other AI-powered chats, e.g. ChatGPT. They find the chat's default
sustainability persona in its current form to be intrusive and patronising.
Methodological limitations
This study faces some methodological limitations. The heterogeneity of the focus groups
enabled us to identify group-specific aspects and to compare the findings of each group with the
other groups. However, with a group size of six participants each, the findings cannot be
generalised for the population and we found some inconsistencies when analysing the feedback
on the features we tested. Furthermore, our research is limited and based solely on perceptions
within the German population. We recommend verifying the group comparisons in conjunction
with quantitative studies. In addition, the influences in focus groups discussed in the
methodological literature, such as response bias, power imbalance within the group and social
desirability, also influence the transferability of our findings. Focus group research can lead to
socially accepted opinions dominating the discussion and certain types of participants taking
over the research process. It can also be difficult to analyse the data and distinguish between
individual and group opinions. To avoid some of these limitations, we carefully assembled and
moderated the groups. Furthermore, we did not attribute opinions to individual participants or
to the group as a whole, but considered them as discourses that emerged in the context we were
studying.
Practical implications
The qualitative feedback from the different groups on Ecosia’s sustainability features helps to
understand the motivations and opinions of potential users. This in turn supports the design and
implementation of effective interventions to change behavior in the area of sustainable
consumption. Overall, we suggest carefully implementing and further testing all the sustainability
features presented, as each of them was found useful by at least one (or more) of the six focus
groups. However, several practical implications emerge from the results of this study.
First, since users often prioritise functional aspects over environmental concerns when using
search engines, sustainability features integrated into search engines should not interfere with
users' primary search activities, but rather complement them seamlessly. Secondly, users expect
a certain autonomy in their search behavior for sustainable information. Therefore, sustainability
interventions should be designed to minimise perceived behavioral impositions or interference
with their search routines, respect users' self-determination process towards environmental
behavior, and avoid any sense of coercion or paternalism. Thirdly, we suggest improving the user
relevance of the features. For example, the name of Ecosia Chat could be changed to avoid
confusion about its functionality. Allowing users to choose from more than just the default chat
character for sustainability would likely increase their interest in using an AI-powered chat
feature. Also, a participatory approach to the sustainability tips and eco-quiz could increase
relevance by allowing users to suggest relevant topics or specific content that Ecosia could
curate. Fourth, different user groups have varying levels of environmental awareness and
attitudes. Tailoring sustainability features to cater to the specific needs and preferences of each
group is essential for engagement and effectiveness. For example, while some groups respond
28
positively to daily sustainability tips, others find them unnecessary or even irritating. Therefore,
we recommend applying target group-specific segmentation strategies to effectively meet the
needs of potential users. The segmentation of user personas could be based on socio-
demographic and psychographic characteristics such as environmental attitudes, climate
awareness, environmental behavior, willingness to change and could include trigger points that
influence their (positive or negative) attitude towards sustainability interventions. On the other
hand, the targeted implementation of sustainability features for specific user groups may require
the collection of user data and the implementation of intention recognition models - which
Ecosia has not yet planned for data protection reasons. However, developing a long-term, value-
adding relationship with users is an important asset of sustainability marketing (Belz & Peattie,
2010). This is also reflected in some of the discussions in the focus groups, where it was
emphasised that the features presented in their current form are most attractive to users who
already have a strong sustainability orientation. While it is generally difficult for marketers of
sustainability products to differentiate between target groups in order to provide users with a
differentiated offering that meets their needs, Ecosia can develop more relevant user features
through qualitative user segmentation via interview studies and gaining deep insights into user
needs, preferences and behaviors without compromising user privacy. Finally, continuous
feedback and iteration based on user reactions are crucial for refining sustainability features and
ensuring their effectiveness. Both positive and negative user feedback can help improve the
design and implementation of sustainability features in search engines.
In terms of their effectiveness and suitability as behavior change interventions (see
Introduction), the presented sustainability features generally meet the requirements for such
techniques. According to the focus groups findings, the sustainability tips provide basic
information and enable practical knowledge to be gained about daily consumption decisions in
favour of sustainability. Overall, they have proven suitable as a ‘knowledge shaping’ technique and
perform best in terms of user interest. However, it became clear that the personal assessment
of the relevance and innovative nature of the information provided is decisive for the willingness
to use the feature in real life. Additionally, it is crucial to consider the social constraints of users
in the tone of voice of these sustainability communication approaches when targeting
individuals' sustainability behavior. For Ecosia, the individualization demanded by the
participants remains a challenge, as it has no detailed information about individual user interests
or the previous search history of individual users. The intervention element dedicated to the
‘behavioral comparison' technique, the tips combined with a poll, did not prove to be effective or
attractive in the eyes of the focus group participants. While in most groups the poll was
described as creating a sense of social pressure and frustration, only the Open-minded found the
social comparison with others encouraging to make behavioral changes. Although the eco-quiz
is perceived in some groups as a fun activity with a learning effect, the possibility of receiving a
‘social reward’ for choosing the right option was not mentioned at all in the groups. Instead, the
eco-quiz is discussed as a playful and informative function with little potential for constant use.
To conclude, the focus groups showed that the perception of Ecosia’s sustainability features
discussed with participants of different environmental awareness groups is mixed. If they are
perceived as unattractive and disturbing or as helpful and attractive highly depends on
contextual and individual factors such as the personal relevance of the content, the general
timing, and the mood in which the user is confronted with the sustainability feature. Regardless
of the technique or the feature itself, it was shown that participants are concerned by possible
29
paternalism of such interventions. If participants feel moralised, they refuse to use the feature.
On the other hand, new technologies such as AI-based conversational agents trigger polarised
and nuanced discussions including enthusiasm, at least in tech-savvy user groups. Overall, the
qualitative study underlines the importance of user segmentation and targeting to improve the
effectiveness of interventions - and eventually to achieve sustainable behavior change apart
from already environmental conscious user groups. Integrating sustainability features into
search engines therefore requires a nuanced understanding of users' preferences, behavior and
attitudes towards sustainability. By addressing users' concerns, respecting their autonomy and
providing relevant and credible information, search engines can play an important role in
promoting more sustainable behaviors and lifestyles.
REFERENCES
Belz, F.M., & Peattie, K. (2010). Sustainability Marketing. A Global Perspective. Journal Consum
Policy 33, 425–426. https://doi.org/10.1007/s10603-010-9132-7
Bohnsack, R. (1989). Generation, Milieu und Geschlecht – Ergebnisse aus Gruppendiskussionen
mit Jugendlichen. Verlag für Sozialwissenschaften Wiesbaden.
https://doi.org/10.1007/978-3-322-97196-8
Calder, B. J. (1977). Focus Groups and the Nature of Qualitative Marketing Research. Journal of
Marketing Research, 14(3), 353-364. https://doi.org/10.2307/3150774
Federal Ministry of the Environment & German Environmental Agency (2023).
Umweltbewusstsein in Deutschland 2022: Ergebnisse einer repräsentativen
Bevölkerungsumfrage, Berlin, retrieved January 17, 2024, from
https://www.umweltbundesamt.de/sites/default/files/medien/3521/publikationen/um
weltbewusstsein_2022_bf-2023_09_04.pdf
GlobeScan (2023). Healthy & Sustainable Living Report 2023, retrieved January 17, 2024, from
https://globescan.wpenginepowered.com/wp-
content/uploads/2023/10/GlobeScan_Healthy_and_Sustainable_Living_Highlights_Rep
ort_2023.pdf
Henseling, C., Hahn, T. & Nolting, K. (2006). Fokusgruppen-Methode als Instrument in der
Umwelt- und Nachhaltigkeitsforschung. WerkstattBericht Nr. 82, retrieved January 17,
2024, from https://nbn-resolving.de/urn:nbn:de:kobv:109-opus-13308
Hyden, L. C., & Bulow, P. H. (2003). Who’s talking: Drawing conclusions from focus groups-
some methodological considerations. International Journal of Social Research
Methodology, 6(4), 305–321. https://doi.org/10.1080/13645570210124865
Krueger, R. A., & Casey, M. A. (2014). Focus groups: A practical guide for applied research. Sage
Publications.
Lewandowski, D. (2012). Credibility in Web Search Engines. In S. Apostel & M. Folk (eds.) (2012).
Online Credibility and Digital Ethos: Evaluating Computer-Mediated Communication (pp.
131–146). IGI Global. https://doi.org/10.4018/978-1-4666-2663-8.ch008
Mau, S., Lux, T. & Westheuser, L. (2023). Triggerpunkte: Konsens und Konflikt in der
Gegenwartsgesellschaft. Suhrkamp.
30
Mayring, P. (2000). Qualitative content analysis. Forum Qualitative Social Research, 1(2).
http://dx.doi.org/10.17169/fqs-1.2.1089
Michie, S., van Stralen, M.M. & West, R. (2011). The behaviour change wheel: A new method for
characterising and designing behaviour change interventions. Implementation Science,
6(1). https://doi.org/10.1186/1748-5908-6-42
Michie, S., Richardson, M., Johnston, M., Abraham, C., Francis, J. J., Hardeman, W., Eccles, M.,
Cane, J. E., & Wood, C. (2013). The Behavior Change Technique Taxonomy (v1) of 93
Hierarchically Clustered Techniques: Building an International Consensus for the
Reporting of Behavior Change Interventions. Annals of Behavioral Medicine, 46(1), 81–95.
https://doi.org/10.1007/s12160-013-9486-6
Pocock T., Smith M. & Wiles J. (2021). Recommendations for Virtual Qualitative Health Research
During a Pandemic. Qualitative Health Research, 31(13), 2403-2413.
https://doi.org/10.1177/10497323211036891
Rosenberg, B. D. & Siegel, J. T. (2018). A 50-Year Review of Psychological Reactance Theory: Do
Not Read This Article. Motivation Science 4(4), 281–300.
https://doi.org/10.1037/mot0000091
Schulz, M., Mack, B. & Renn, O. (2012): Fokusgruppen in der empirischen Sozialwissenschaft.
Von der Konzeption bis zur Auswertung. VS Verlag für Sozialwissenschaften Wiesbaden.
https://doi.org/10.1007/978-3-531-19397-7
Stieß, I., Sunderer, G., Raschewski, L., Stein, M., Götz, K., Belz, J., Follmer, R., Hölscher, J.,
Birzle-Harder, B. (2022). Repräsentativumfrage zum Umweltbewusstsein und
Umweltverhalten im Jahr 2020. Klimaschutz und sozial-ökologische Transformation.
UBA-Texte 20/2022. Dessau-Roßlau, retrieved January 17, 2024, from
https://www.umweltbundesamt.de/publikationen/repraesentativumfrage-
umweltbewusstsein-0
Strauss, A., & Corbin, J. (1998). Basics of Qualitative Research: Techniques and Procedures for
Developing Grounded Theory. Sage Publications.
The Future Shopper Report (2022). Wunderman Thompson Commerce, retrieved January 17,
2024, from https://www.wundermanthompson.com/insight/the-future-shopper-2022
Warr, D. J. (2005). ‘It was fun… but we don’t usually talk about these things’: Analyzing Sociable
Interaction in Focus Groups. Qualitative Inquiry 11(2), 200-225.
https://doi.org/10.1177/1077800404273412
