Exploring travel agencies customers’ loyalty motives throught machine learning analysis

Ma ke ing and T ade 201 2024, olume 27, issue 2, pp. 201–220, DOI: 10.15240/ ul/001/2024-2-012 Explo ing a el agencies cus ome s’ loyal y mo i es h ough machine lea ning analysis And zej Dudek1, Da ia E. Ja emen2, Izabela Michalska-Dudek3, Pa lina Pelleso a4 1 W oclaw Uni e si y o Economics and Business, Facul y o Economics and Finance, Depa men o Econome ics and Compu e Science, Poland, ORCID: 0000-0002-4943-8703, [email p o ec ed]; 2 W oclaw Uni e si y o Economics and Business, Facul y o Managemen and Business, Depa men o Ma ke ing and Tou ism Managemen , Poland, ORCID: 0000-0001-8726-7573, [email p o ec ed]; 3 W oclaw Uni e si y o Economics and Business, Facul y o Managemen and Business, Depa men o Ma ke ing and Tou ism Managemen , Poland, ORCID: 0000-0003-2804-6513, [email p o ec ed] (co esponding au ho ); 4 Silesian Uni e si y in Opa a, School o Business Adminis a ion in Ka ina, Depa men o Tou ism and Leisu e Ac i i ies, Czech Republic, ORCID: 0000-0002-2595-923X, [email p o ec ed]. Abs ac : This a icle ocuses on iden i ying he di ec ions o changes in he decision-making p ocess o pu chasing package a els and he mo i es de e mining his pu chase, as well as he impac o hese mo i es on he a ec i e, beha io al and global loyal y o a el agency cus ome s du ing he COVID-19 pandemic. To achie e he esea ch goal, a li e a u e e iew and ield esea ch we e conduc ed. In he case o seconda y sou ces, he con en analysis me hod was used o examine hem, while da a om p ima y sou ces (N = 1,508) we e collec ed using an indi ec su ey echnique (CAWI). The da a analysis was ca ied ou using machine lea ning – he a iable impo ance me hod and he andom o es algo i hm. The ob ained esul s allowed us o conclude ha du ing he pe iod o h ea o he COVID-19 i us pandemic, ou is packages had been pu chased less equen ly, and buye s ei he had limi ed hei ips o domes ic ips o had adap ed hei a el des ina ions o he changing pandemic si ua ion. The mos impo an mo i a o s ha de e mined he choice o a a el agency du ing he COVID-19 pandemic we e an a ac i e o e , a wide selec ion o package a els, p e ious posi i e expe iences and us in he o ganise . I was also con i med ha in he ace o he pandemic h ea , buye s o package a els had been loyal o a el agencies. The use o machine lea ning allowed o mo e in-dep h analyses and iden i ica ion o mo i es ha had a key impac on he de elopmen o buye loyal y du ing he pandemic. The ac o s iden i ied in he s udy encou aging buye s o package a els o main ain long- e m ela ionships wi h hei supplie s a e belie in he alue o he a el agency’s o e , us in he a el agency, indi idual app oach a el agency employees, e icien se ice in he a el agency, and a wide selec ion o package a els. Keywo ds: Tou is package, a el agency, cus ome beha io , andom o es , COVID-19 pandemic. JEL Classi ica ion: D12, Z31, Z33. APA S yle Ci a ion: Dudek, A., Ja emen, D. E., Michalska-Dudek, I., & Pelleso a, P. (2024). Explo ing a el agencies cus ome s’ loyal y mo i es h ough machine lea ning analysis. E&M Economics and Managemen , 27(2), 201–220. h ps://doi.o g/10.15240/ ul/001/2024-2-012 In oduc ion The sale o o ganised ou is ips o e ed by ou ope a o s cons i u es one hi d o global ou ism e enues (S a is a, 2021). These ips a e called many names, e.g., ou is packages, pack- age holidays, ou is se ices packages, and Ma ke ing and T ade 202 2024, olume 27, issue 2, pp. 201–220, DOI: 10.15240/ ul/001/2024-2-012 package a els (acco ding o Di ec i e (EU) 2015/2302 o he Eu opean Pa liamen and o he Council o No embe 25, 2015, on pack- age a el and linked a el a angemen s, amending Regula ion (EC) No. 2006/2004 and Di ec i e 2011/83/EU o he Eu opean Pa lia- men and o he Council and epealing Council Di ec i e 90/314/EEC). Simila ly o he en i e ou ism indus y, a el agencies ha e also been a ec ed by he COVID-19 pandemic in 2020–2022. As a esul , in he i s yea o he pandemic (2020) compa ed o 2019, sales o package a els dec eased by 64%. Tou is s ga e up a eling mos o en unde he p es- su e o legal es ic ions p ohibi ing mo e- men and c ossing s a e bo de s. The e ec o educed demand o in e na ional ips was ob ious and easies o obse e and es ima e. Many esea che s, howe e , assumed ha in addi ion o he o ced cancella ion o a el, he pandemic had had a much b oade impac on he beha io o ou is s, including buye s o package a els. All he mo e so because in o ma ion published by he in e ne pla o m Wasza. u ys yka.pl in 2021 om in e iews wi h ep esen a i es o ou ope a o s and a el agen s, as well as p i a e con e sa ions o he au ho s o he a icle wi h p ac i ione s, signaled ha he impac o he pandemic had no been limi ed o he decline in he numbe o o ganised ips, bu also had a ec ed o he aspec s o a eling (dis ance, leng h o s ay, he me hod o anspo , accommoda ion con- di ions) and on he decision-making p ocess i sel o pu chase package a el and he ex- pec a ions exp essed by cus ome s owa ds he se ice in a el agencies (Wasza u ys yka, 2020, 2021). Fo his eason, he au ho s con- cluded ha a mo e ho ough unde s anding o he beha io o buye s o package a els was in e es ing bo h o scien i ic easons and may ha e had a u ili a ian alue o ep esen a- i es o a el agencies. The main aim o he a icle was o iden i y he di ec ions o changes in he decision-making p ocess o pu chasing package a el and he mo i es de e mining his pu chase, as well as he impac o hese mo i es on he a ec i e, beha io al and global loyal y o a el agency cus ome s du ing he COVID-19 pandemic. The e a e ou esea ch ques ions asked: RQ1: Was he decision-making p ocess o pu chasing package a els du ing he COVID-19 h ea di e en om be o e he pandemic? RQ2: Wha mo i a o s de e mined he choice o a package a el supplie ( ou ope a- o ) du ing he COVID-19 pandemic? RQ3: We e pu chase s o package a els loyal o a el agencies du ing he COVID-19 pandemic h ea ? RQ4: Did he iden i ied easons o choos- ing a package a el supplie in luence he loy- al y o a el agency cus ome s? The da a needed o sol e he esea ch p oblem unde aken in he a icle was collec ed om bo h seconda y and p ima y sou ces. In he o - he-desk esea ch, he me hod o sys- ema ic con en analysis o seconda y sou ces was used. In he case o p ima y esea ch, da a was ob ained using a selec ed indi ec su ey me hod, namely a compu e -assis ed web in e iew. 1. Theo e ical backg ound 1.1 Loyal y o package a el buye s Cus ome loyal y is one o he mos impo an long- e m goals o all businesses, including ou ope a o s. Tou ope a o s’ bene i s om buye loyal y a e clea . Cus ome s make e- pea pu chases o a pa icula ou ope a o ’s se ice packages and, a he same ime, hey adop an a i ude a o able o he p o ide and he b and i o e s (in ou ism, he b and name is usually iden ical o ha o he ou ope a- o ). Loyal buye s e y o en s ep in o he ole o a b and gua dian, an ambassado and e en a de ende , which has posi i e conno a- ions, especially in imes o c isis. Indeed, as Rundle-Thiele (2006) no es, loyal y is a buy- e ’s commi men o pe sis in a ela ionship wi h a supplie o be e o wo se. An excep ionally loyal cus ome does no se iously conside al- e na i e b ands and, mo eo e , o en igno es o he supplie s (De ece e al., 2015). The b and o which he buye is loyal has a leading posi- ion and appea s i s in his o he mind when making a pu chase decision (Han e al., 2011), hus achie ing loyal y a he cogni i e le el (Ca uana, 2002). Some esea che s, such as Reichheld and Sche e (2000), a gue ha , om he cus ome ’s pe spec i e, b and loyal y is a mo e impo an ac o in he choice o sup- plie han p ice, while Chaudhu i and Holb ook (2001) and Rowley (2005) add ha loyal y e- duces he p ice sensi i i y o he buye . The e m “loyal” can be unde s ood in many ways (Mo gan, 1999), and i s de ini ion has Ma ke ing and T ade 203 2024, olume 27, issue 2, pp. 201–220, DOI: 10.15240/ ul/001/2024-2-012 e ol ed o e ime (Campo & Yagüe, 2008). In he ma ke ing li e a u e, cus ome loyal y is usually de ined as ei he beha iou s (Newman & We bel, 1973) o a i udes owa ds a b and (Bendapudi & Be y, 1997) o a mix u e o bo h dimensions (Day, 1969; Dick & Baśu, 1994; Goldbe g, 1981; G emle & B own, 1996; Jaco- by & Ches nu , 1978; Oli e , 1999). In he i s sense, known as beha iou al, loyal y o a ou ope a o e e s o a buye ’s beha iou o e- pea edly buying package a el om he same b and (e.g., Gonçal es & Sampaio, 2012). Loyal y in e p e ed in his way is p ima ily ex- p essed in he equency and olume ( alue) o pu chases, which ansla es in o he sup- plie ’s ma ke sha e (Richa d & Zhang, 2012). Howe e , i does no answe he ques ion o he easons and mo i es explaining his epea pu chase (Dick & Baśu, 1994). In con- as , he a ionale o loyal y, a leas he psy- chog aphic one, is con ained in he a ec i e app oach. In his iew, loyal y is he esul o buye s’ emo ions associa ed wi h a ou op- e a o ’s b and, buye s’ subjec i e eelings and hei a i ude owa ds he b and, which is ex- p essed in conscious a achmen o he b and and a desi e o con inue a ansac ional ela- ionship wi h a pa icula supplie (Chen & Tsai, 2008). A ec i e loyal y also includes buye s’ p e e ences, us in he supplie , commi men o he ela ionship (De ece e al., 2015) and ecommenda ion o he b and o o he po en ial buye s (Konecnik & Ga ne , 2007). The li e a- u e sou ces emphasise ha loyal y is a unc- ion o hough (men al) p ocesses in ol ing e alua ion ( alua ion) and he buye ’s decision based on in e nal conside a ions (And eas- sen & Lindes ad,1998). Taking in o accoun he wo dimensions o loyal y men ioned abo e, Oli e (1999) s esses ha loyal y exp esses he deep- oo ed obliga ion o he buye o e- pu chase o ecommend he b and in he u- u e despi e he in luence o en i onmen al (si ua ional) ac o s and he ma ke ing e o s o supplie s. Many o he esea che s p esen a mixed app oach o he concep o loyal y. Among hem, G emle and B own (1996) claim ha he loyal y o a se ice buye is he deg ee o which he cus ome epea s pu chasing be- ha io wi h he se ice p o ide has a posi i e a i ude owa ds him and conside s using only his supplie when he need o a gi en se ice a ises ( o a ou is package). These combined dimensions yield wha is known as global (mixed) loyal y, which is he p oduc o a i udes and ac ions (Dick & Basu, 1994; Oli e , 1999). In a s uc u ed way, global loyal y is de ined by And eassen and Lindes ad (1998), who s a e ha i is a non- andom (subjec i e), beha iou al esponse (i.e., pu chase), exp essed o e ime by a decision-making indi idual, o one o mo e al e na i e b ands among a se o such b ands. I should also be men ioned he e ha ecen ly, in he case o ou is packages, buye loyal y is inc easingly being conside ed in he con ex o he In e ne , because he ac i i ies o a el agencies a e unde going dynamic digi isa ion. S a is a (2021) epo s ha 70% o sales o o - ganised ou is ips a e sold in e-comme ce. The loyal y o online a el package buye s esul s om hei a o able a i ude owa ds on- line shopping and epea pu chases in he on- line channel (Ande son & S ini asan, 2003). In ac , unde s anding he abo e-men ioned loyal y pe spec i es is impo an o he con- cep ualisa ion o esea ch and he cons uc- ion o an ins umen measu ing cus ome loyal y, because each o hese in e p e a ions has di e en consequences o he assess- men o mo i es and ways o measu ing loyal y (Gonçal es & Sampaio, 2012; Mazu ek-Kusiak, 2020). Beha io al loyal y is usually measu ed by he equency o pu chases. Al hough, ac- co ding o De ece e al. (2015), in he case o pu chases o ou is packages, his is no an ideal measu e due o he ac ha , in es- sence, he packages a e pu chased a he spo adically (once o wice a yea ), no o en, which is he esul o hei high cos and s ic dependency on he use om he package om he clien ’s ee ime (holiday), which is usually limi ed. A be e measu e would, he e o e, be epea pu chases. An analysis o he li e a- u e p o ides impo an insigh ha is a end in he s udy o beha io al loyal y, ocusing on he le el o cus ome in en ion/ eadiness o epea he pu chase (Ama o & Dua e, 2015; Del Bosque e al., 2006). The s udy o a ec- i e loyal y in ol es measu ing buye a i udes, which may esul in epea pu chases. This mea- su emen is much mo e complex and includes, among o he s: measu ing sa is ac ion, b and p e e ence o e o he b ands, and he cus- ome ’s endency o ecommend he supplie o o he s. In his app oach, he measu e may be he p obabili y ha cus ome s will exp ess posi i e opinions abou he a el agency, ecommend i o o he s and encou age o he s Ma ke ing and T ade 204 2024, olume 27, issue 2, pp. 201–220, DOI: 10.15240/ ul/001/2024-2-012 o pu chase he se ices o his agency, and he p obabili y ha cus ome s will conside a gi en ou ope a o as he supplie o i s choice when hey ha e a need o use a ou is package (Richa d & Zhang, 2012). In s udies on ou is s’ loyal y, an impo an ole also play he ac o s ha encou age people o be loyal, i.e., he de e minan s o loyal y. In he case o hese s udies, loyal y ac o s o ou ism package buye s. The analysis o he achie emen s o p e ious esea che s on his issue, including 30 ele an a icles, delibe a ely selec ed om he Scopus and Web o Science da abases, published o e he las De e minan s o loyal y Resea che s Sa is ac ion Del Bosque e al. (2006), Lee e al. (2011), Se ó-Pamies (2012), Majo and McLeay (2013), P ause e al. (2013), Lai (2014), Sil a and Gonçal es (2016), Yang e al. (2017), Rizal e al. (2020), Ce in (2020), A siz e al. (2021), Mu sid and Wu (2022), San os e al. (2022) Pe cei ed alue o se ices/ ela ional bene i s Conze e al. (2010), Sende s e al. (2013), Lai (2014), Roge -Monzó e al. (2015), Sil a and Gonçal es (2016), Ronsana e al. (2018), Dudek e al. (2019), Albay ak e al. (2020), Mu sid and Wu (2022) Pe cei ed quali y o se ice/ ela ionships/ communica ion quali y o websi e/sys em Lee and Cunningham (2001), Campo and Yagüe (2008), Huang (2008), Thi Phuong Thao and Swie czek (2008), Lee e al. (2011), Se ó-Pamies (2012), Majo and McLeay (2013), Lai (2014), Moisescu and Gica (2014), Roge -Monzó e al. (2015), De ece e al. (2015), Sil a and Gonçal es (2016), Abou-Shouk and Khali a (2017), Dudek e al. (2019), Albay ak e al. (2020), Rizal e al. (2020), Ce in (2020), Elga hy (2022) T us in he supplie Huang (2008), Se ó-Pamies (2012), P ause e al. (2013), Lai (2014), De ece e al. (2015), Sil a and Gonçal es (2016), Ronsana e al. (2018), Albay ak e al. (2020), Rizal e al. (2020), San os e al. (2022) P ices/ ansac ion cos s Lee and Cunningham (2001), Campo and Yagüe (2008), Thi Phuong Thao and Swie czek (2008), Sende s e al. (2013), Elga hy (2022) Emo ional in ol emen o cus ome s/cus ome iden i ica ion wi h he company Richa d and Zhang (2012), P ause e al. (2013), Lai (2014), Yang e al. (2017), Elga hy (2022) P o ide ’s b and and i s image Richa d and Zhang (2012), Dudek e al. (2019) Co po a e social esponsibili y (CSR) Moisescu and Gica (2014), Akba i e al. (2020) Expe ience and i s dimensions Ce in (2020) Expec a ions Del Bosque e al. (2006), Lee e al. (2011) Mo i a ions/goals o a elling Lee e al. (2011), Mody e al. (2017) Pe cep ion o b and Mody e al. (2017) Cus ome o ien a ion Yang e al. (2017) Complain s Lee e al. (2011) Loyal y p og ammes Elga hy (2022) Sou ce: own Tab. 1: De e minan s o a el agency cus ome loyal y conside ed in p e ious s udies Ma ke ing and T ade 205 2024, olume 27, issue 2, pp. 201–220, DOI: 10.15240/ ul/001/2024-2-012 h ee decades, p o ides many aluable ips o s udying he de e minan s o loyal y (Tab. 1). Fi s ly, he mos equen ly men ioned mo i a- o s o being a loyal cus ome a e he pe - cei ed quali y and pe cei ed alue o se ices, sa is ac ion, and secondly, us in he supplie , p ice o se ices, emo ional in ol emen , sup- plie ’s image, as well as co po a e social e- sponsibili y and buye ’s expe ience. Indi idual s udies analysed many o he loyal y ac o s, e.g., loyal y p og ams, complain handling, a el mo i a ions and consume expec a ions. I should be no ed ha a sys ema ic in- c ease in he numbe o s udies conduc ed in he a ea o ac o s shaping loyal y ocuses on he de e minan s o buye s’ loyal y o a el agencies (agen s and ou ope a o s) ope a ing in an e-comme ce en i onmen (Abou-Shouk & Khali a, 2017; Albay ak e al., 2020; Huang, 2008; Rizal e al., 2020; Ronsana e al., 2018; Sil a & Gonçal es, 2016). Thei esul s clea ly poin o new ac o s speci ic o he online en- i onmen , such as buye s’ in ol emen in so- cial media, websi e design, quali y o se ice in unc ional and hedonic e ms, in e ac ional and ansac ional bene i s o he cus ome , as well as online in e ac ions wi h o he cu en and po en ial buye s. The abo e li e a u e e iew om wo highly a ed da abases o scien i ic publica- ions indica es ha o e he las 30 yea s, only 30 a icles ha e appea ed on he issues o he d i e s o cus ome loyal y in a el agencies. This means ha , on a e age, 1 a - icle ela ed hema ically was published in one yea and his is no an imp essi e numbe . I should be admi ed ha in he pe iod unde s udy, i is possible o indica e yea s in which he in e es in he men ioned p oblem was g ea e (3–4 a icles in 2017, 2020 and 2022), bu i passed ela i ely quickly, and no clea ly dominan di ec ions o esea ch on he loyal y o buye s o package a els eme ged. In iew o he de elopmen o new in o ma ion and communica ion echnologies and he eme - gence wi h hem o he unlimi ed possibili y o indi idual a elle s o o ganise hei own ips (bypassing he in e media ies), cus ome e en ion by package a el supplie s ( a el agencies) is now becoming a mo e impo an ask han i was be o e he In e ne e olu ion. And he limi ed esea ch so a , o en selec- i ely ea ing he p oblem o cus ome loyal y ac o s o a el agencies, con i ms he need o conduc u he esea ch and a mo e com- p ehensi e app oach o iden i ying and o ganis- ing hese de e minan s. Addi ionally, he analysis o he de e mi- nan s o a el agency cus ome ’s loyal y in ligh o p e ious esea ch has allowed he inal concep ualisa ion o he esea ch p oblem ad- d essed in his s udy, as well as he ope a ion- alisa ion o he esea ch and he cons uc ion o an app op ia e measu emen ins umen ha akes in o accoun he scien i ic achie emen s o o he esea che s. 1.2 COVID-19 and he loyal y o buye s o package a els in he ligh o p e ious esea ch In o de o con i m he esea ch gap ega ding he issue o shaping he loyal y o buye s o ou - is se ice packages in a pandemic si ua ion, a sys ema ic e iew o he li e a u e on he impac o COVID-19 on he beha io o a el agency cus ome s was conduc ed. In he p ocess o se- lec ing a icles accep ed o he e iew, he da- abases o scien i ic publica ions mos alued by he academic communi y we e used i s , i.e., Web o Science (WoS) and Scopus, and hen he lis o a icles was supplemen ed wi h publica- ions ha we e p esen ed by he Google Schola sea ch engine on he 50 pages exposed i s . The selec ion o li e a u e was done acco ding o a speci ic ca ego isa ion key and an assumed il e . The ollowing selec ion c i e ia we e used: pe iod o publica ion (2020–2023), language o publica ion (English), access ( ull- ex ) and, ype o publica ion (a icle in a scien i ic jou nal o chap e (s) in a scien i ic monog aph) and na u e o publica ion ( esea ch o e iew). The ca ego i- sa ion key na owed he sea ch o i les, abs ac s and keywo ds o publica ions. Da abases we e b owsed by en e ing he app op ia e combina ion o he ollowing keywo ds: “loyal y,” “packaged ou (s),” “package a el(s),” “ ou is package(s),” “holiday package(s),” “ a el agency(s),” “ ou ope a o (s)” and “COVID-19” (Tab. 2). A sea ch o he da abases yielded esul s, including a ew scien i ic publica ions (Sco- pus – 4, and Web o Science – 11) ouching on he issue o package a el buye s’ loy- al y unde he COVID-19 pandemic. A ca e ul analysis o he con en o hese publica ions u he e ealed ha only wo o hem di ec ly add essed he impac o he pandemic on a el agency cus ome loyal y. The i s a icle o el- e ance, by Wu and Rian ama (2022), discusses Ma ke ing and T ade 206 2024, olume 27, issue 2, pp. 201–220, DOI: 10.15240/ ul/001/2024-2-012 he p oblem o he eme gence o new p edic o s o sa is ac ion – conside ed one o he main ac- o s mo i a ing buye s o be loyal. The esea ch- e s ound ou ha such a new and ele an mo i e o loyal y in an eme gency si ua ion was he a el agen ’s eac ion o he occu - ence o an unexpec ed e en ha could dis up he pe o mance o he objec o he ansac ion. The ision o losing he money paid o a sup- plie o a p epaid package a el long be o e he se ice is p o ided o being unable o use he pu chased package we e he bigges ea s o ou is s du ing he pandemic. This ea o en demo i a ed ou is s om unde aking ou is ips du ing he pandemic. On he o he hand, Zhu e al. (2022) conside ed he ela ionship be ween he eliabili y o a el agency websi es and cus ome loyal y. The esul s o hei s udy indica ed ha du ing he pandemic pe iod, he con en p esen ed on a supplie ’s websi e did no ha e a signi ican impac on buye s’ a i- udes and pu chase in en ions. This is because he unce ain y o he en i onmen and he dy- namics o he si ua ion du ing he COVID-19 pe iod caused his con en o quickly become ou da ed, and as a esul , cus ome s did no pay much a en ion o i . A ce ain pauci y o esul s le a e e iew- ing he Scopus and Web o Science da abases p omp ed he au ho s o expand hei sea ch o o he sou ces o in o ma ion. Fo he pene a- ion, Google Schola was chosen (as he mos popula b owse o schola ly esou ces oday), al hough no wi hou ce ain impe ec ions, bu con aining a e y wide selec ion o s udies. This addi ional e iew made i possible o supplemen he wo ea lie eco ds wi h wo mo e. In hei s udy, Lawu e al. (2022), simila ly o Wu and Rian ama (2022), included among he de- e minan s o a el agency cus ome loyal y he esponse o agencies o possible dis up ions in he deli e y o he pu chased a el se ice. The esea che s ound ou ha cus ome loyal y du ing COVID-19 was in luenced by so-called p ocedu al ai ness, i.e., he ways (p ocedu es) a supplie handles p oblems wi h he deli e y o a se ice. Implemen ing app op ia e se ice s anda ds in a se ice cancella ion si ua ion, on he one hand, builds a sense o secu i y and us in he supplie , and on he o he hand, inc eases cus ome sa is ac ion. In con as , he esul s o a s udy by Jas o ia e al. (2023) indica ed an in- c ease in he impo ance o expe ience-based us in building loyal y among online a el agency cus ome s du ing he pandemic. Be o e he pandemic, us was no as signi ican a p e- dic o o loyal y in en ions. The pandemic si ua ion – a comple ely new, p e iously unknown condi ion o conduc - ing business in ou ism is an a ac i e es ing g ound and an in e es ing ield o es ing known esea ch p oblems, bu in he con ex o a new en i onmen . New condi ions may b ing new insigh s and change cu en hinking abou loy- al y a iables, hey may also con i m p e ious indings, and hei esul s will no be su p ising. De e mining whe he pandemic condi ions ha e al e ed loyal y ac o s owa ds he choice o a el agen s wa an s u he esea ch. The au ho s o his s udy seek o ill his gap. 2. Resea ch me hodology 2.1 Desc ip ion o he s udy and cha ac e is ics o he sample In o de o achie e he es ablished esea ch goals, buye esea ch was ca ied ou on Combina ion Numbe o pape s Scopus Web od Science Loyal y AND packaged AND ou AND COVID-19 – – Loyal y AND ou is AND package AND COVID-19 – – Loyal y AND holiday AND package AND COVID-19 – 1 Loyal y AND package AND a el AND COVID-19 – 1 Loyal y AND ou is / a el AND agency AND COVID-19 3 9 Loyal y AND ou AND ope a o AND COVID-19 1 – Sou ce: own Tab. 2: Quan i a i e esul s o he li e a u e e iew Ma ke ing and T ade 207 2024, olume 27, issue 2, pp. 201–220, DOI: 10.15240/ ul/001/2024-2-012 he issue o a eling du ing he COVID-19 pan- demic. In he conduc ed su ey, he basic ool was a ques ionnai e con aining 38 ques ions. The esea ch was a pa ial s udy and was ca - ied ou using an online su ey on a na ionwide online panel o esponden s using he CAWI me hod. The s udy used a non-p obabilis ic sample selec ion – he In e ne sampling me h- od. The su ey was conduc ed om June 1 o Augus 31, 2022, and ul ima ely included a sample o 1,508 esponden s (N = 1,508). The su eyed Poles (N = 1,508) included ep esen a i es o bo h sexes, and he mos nume ous age g oups we e om 36 o 45, om 26 o 35, and om 46 o 60 yea s o age (31.0%, espec i ely 27.90% and 19.2% o e- sponden s). The majo i y o esponden s we e ma ied (58.2%) and esponden s wi h highe educa ion (50.7%). Responden s’ households usually consis ed o 2 o 4 people (79.8% o e- sponses in o al), and e e y second esponden had unde age child en in hei household. O he esponden s, 83.3% we e wo king peo- ple. The la ges g oup o esponden s (75.7%) we e esponden s li ing in ci ies. Nea ly hal o he esponden s assessed hei inancial si u- a ion as good, and only less han 1.3% o hem hough i was e y bad. The s udy was cha ac e ised by a ep- esen a i e dis ibu ion o cha ac e is ics o he gene al popula ion o Poles. Compa ison o cha ac e is ics o he dis ibu ion o me ic ea u es wi h hose a ailable in public s a is ics, cha ac e is ics o he popula ion o adul Poles allows us o conclude ha ha he s uc u e o he sample in e ms o gende , age, educa- ion and size o place o esidence co esponds o he s uc u e o he popula ion o adul Poles. The da a analysis was ca ied ou using desc ip i e s a is ics indica o s and machine lea ning – he a iable impo ance me hod and he andom o es algo i hm using he DATEX package o he R p og am. 3. Resul s and discussion 3.1 Analysis o he impac o he COVID-19 pandemic on he beha io o ou is packages buye s The esea ch esul s clea ly con i m ha he si - ua ion ela ed o he COVID-19 pandemic has in luenced Poles’ ou is ips. This was con- i med by as many as 69.1% o esponden s. A de ailed summa y o he esea ch esul s is p esen ed in Tab. 3. Among he changes made in he me hod o ca ying ou ou is ips, he esponden s mos o en indica ed: limi ing he numbe o ou is ips (39.6% o esponses), ca ying ou only domes ic ou is ips (27.9% o e- sponses) and changing he des ina ion o ou - is ips (22.1% o esponses). The emaining answe s a e p esen ed in Tab. 4. The abo e esul s allow o answe he i s esea ch ques ion and p o ide con i ma ion ha in he ace o he h ea o he COVID-19 pandemic, he beha io o pu chase s o pack- age a els has changed signi ican ly. T a el packages a ha ime we e pu chased less equen ly and buye s limi ed hei ips o domes ic ips. The analysis o he answe s p o ided by he esponden s allowed us o answe he sec- ond esea ch ques ion and iden i y he mos impo an hemes ha had guided he pu chas- e s o package a els when choosing hei o ganise (Tab. 5). Has he COVID-19 pandemic a ec ed you ou is a el? To al n(%) De ini ely yes 480 31.8 Yes 562 37.3 Nei he yes no no 256 17.0 No 131 8.7 De ini ely no 79 5.2 Sou ce: own Tab. 3: The impac o he COVID-19 i us pandemic on esponden s’ ou is ips (N = 1,508) Ma ke ing and T ade 208 2024, olume 27, issue 2, pp. 201–220, DOI: 10.15240/ ul/001/2024-2-012 Buye s’ mo i es o choosing a a el agency (BM; on a scale o 1–7) A e age a ing impo ance o he mo i e BM_1 – in e es ing/a ac i e ou is packages 5.68 BM_2 – a wide selec ion o ou is packages 5.66 BM_3 – sa is ac ion wi h p e iously pu chased o e s 5.77 BM_4 – us 5.72 BM_5 – epu a ion and p es ige 5.30 BM_6 – con enien loca ion 5.29 BM_7 – belie in he alue o he o e 5.53 BM_8 – a o able p ice–quali y a io 5.80 BM_9 – sense o secu i y 5.75 BM_10 – special ea men o me by employees 5.14 BM_11 – se ice by a specially designa ed employee 5.15 BM_12 – a achmen 5.05 BM_13 – o e ed loyal y p og am (e.g., loyal y ca d/club) 5.01 BM_14 – e icien /ad iso y se ice 5.62 BM_15 – indi idual app oach o o ice s a 5.46 BM_16 – a o able paymen e ms 5.69 BM_17 – compe i ions and lo e ies 4.48 Wha changes ha e you made o you ou is ips due o he COVID-19 pandemic? To al n(%) None 55 3.6 Comple e esigna ion om ou is ips 248 16.4 Limi ed o domes ic a el only 421 27.9 Change o a el des ina ion 334 22.1 T a el only by p i a e ca 254 16.8 Gi ing up ho els in a o o mo e p i a e places o s ay (e.g., sepa a e apa men s, ag i ou ism lodgings) 199 13.2 Limi ing he numbe o ou is ips 597 39.6 Sho ening he leng h o ou is ips 202 13.4 Comple e esigna ion om pu chasing ou is packages o ganised by a a el agency 241 16.0 Di e en beha io 24 1.6 No e: The answe s do no add up o 100% because he esponden s had he possibili y o making mul iple choices. Sou ce: own Tab. 5 Mo i es o loyal y o he a el agencies – Pa 1 Tab. 4: Changes in ou is a el caused by he COVID-19 pandemic (N = 1,508) Ma ke ing and T ade 209 2024, olume 27, issue 2, pp. 201–220, DOI: 10.15240/ ul/001/2024-2-012 Among he key mo i es in luencing he choice o a a el o ganise , esponden s in- dica ed: in e es ing/a ac i e ou is packages, wide selec ion o ou is packages, us , sense o secu i y, a o able p ice-quali y a io, a o - able paymen e ms, con enience, e icien ad- iso y se ice and sa is ac ion wi h p e iously pu chased o e s. I can he e o e be s a ed ha cus ome s a e looking o in e es ing des ina ions ha sui hei a el p e e ences, as well as he possibil- i y o a wide selec ion and a ie y o o ganised a el op ions ha hey can ailo o hei indi- idual needs. P e ious pu chasing expe ience also plays an impo an ole. Cus ome s who a e sa is ied wi h p e ious ips a e mo e likely o choose he same a el agency in he u- u e, and posi i e expe iences om p e ious ips can build cus ome loyal y and encou - age hem o e u n o he same a el agency. The c edibili y and epu a ion o he a el agency also ha e a signi ican impac on he pu chasing decisions o buye s o ou - is packages. Cus ome s expec ha a a el agency will p o ide hem wi h a sa e jou ney, bo h du ing he ip and du ing hei s ay he e. Cus ome s op imise hei choices because hey look o o e s ha p opose a good alue o he p ice hey pay. Flexible paymen op ions can also be an impo an ac o o cus ome s, especially when pu chasing a holiday, which is a signi ican expense in he household bud- ge . The con enience o pu chasing a package a el, as well as a simple and in ui i e p ocess o booking and pu chasing ou is ips, is also a ac o ha a ac s cus ome s. Mo eo e , e ec i e and help ul cus ome se ice can be ano he impo an ac o in con incing cus om- e s o choose a gi en a el agency. Unde s anding hese ac o s and adap ing he a el agency’s o e o cus ome needs can con ibu e o inc easing hei sa is ac ion and building loyal y. 3.2 Machine lea ning analysis In he nex pa o he esea ch (in sea ch o he answe o RQ4), machine lea ning was used, he main goal o which is he p ac ical applica ion o a i icial in elligence o c ea e an au oma ic sys em capable o imp o ing i sel based on expe ience (da a) and acqui e new knowledge on his basis. Mos o en, machine lea ning is used o de ec unknown pa e ns in da a and o o mula e ules, and i can be success ully used in consume esea ch o de ec pa e ns o ela ionships and p edic u u e consume beha io , such as making epea pu chases. Taking in o accoun he c i e ia o selec ing he algo i hm, such as: he numbe o obse a- ions, he numbe o analysed a iables and he alues o ca ego ical a iables (numbe o labels) in he inpu da a, he andom o es algo i hm was used in he s udy. The idea o ope a ion o his algo i hm is based on classi ica ion using a g oup o decision ees and he bigges di e ence in he s uc u e o bo h algo i hms is he so-called boo s ap. The algo i hm s a s by building many decision ees (in ou case, he numbe o ees was 50). Fo each ee, a andom sample o obse a ions is selec ed, consis ing o se e al explana o y a iables ( he numbe o a iables in each ee is he second pa ame e de ined by he use ). Buye s’ mo i es o choosing a a el agency (BM; on a scale o 1–7) A e age a ing impo ance o he mo i e BM_18 – p ices, gi s, gadge s 4.54 BM_19 – discoun s. discoun s and o he inancial incen i es 5.45 BM_20 – ad e isemen 4.60 BM_21 – no o he a el agencies 4.12 BM_22 – con enience o pu chase 5.75 BM_23 – no ime 4.38 BM_24 – eluc ance o change 4.41 Sou ce: own Tab. 5 Mo i es o loyal y o he a el agencies – Pa 2 Ma ke ing and T ade 216 2024, olume 27, issue 2, pp. 201–220, DOI: 10.15240/ ul/001/2024-2-012 Packaged ou s we e pu chased less equen - ly, and buye s ei he limi ed hei ips o do- mes ic a el o adap ed a el des ina ions o he changing pandemic si ua ion. The mos impo an TOP10 (ou o 24 a i- ables) mo i a o s ha de e mined he choice o a a el agency du ing he COVID-19 pandemic we e: a o able p ice-quali y a io, sa is ac ion wi h p e iously pu chased o e s, sense o secu i y, con enience o pu chase, us , a ou able paymen e ms, an in e es ing/ a ac i e ou is packages, a wide selec ion o ou is packages, e icien /ad iso y se ice and belie in he alue o he o e . I is no ewo - hy ha wo ac o s un ela ed o he economic e iciency o choice anked ela i ely high in his anking. These a e a sense o secu i y ( he hi d place), which is dependen on us in he se - ice p o ide ( he i h place). I u ned ou ha in he ace o he pan- demic h ea , buye s o package a els decla ed loyal y o a el agencies. The s udy ob ained a high le el o loyal y, bo h beha - io al (exp essed by buye s’ decla ed eadiness o make ano he pu chase o package a el) and a ec i e (exp essed by buye s’ eadiness o ecommend a a el agency). This allows us o alk abou he exis ence o a s ong ela ion- ship be ween a el agencies and he buye s o hei o e s du ing he pandemic h ea , which a he same ime allows he answe o he esea ch ques ion RQ3 (We e pu chas- e s o package a els loyal o a el agencies du ing he COVID-19 pandemic h ea ?) and con i ms ha du ing he h ea o he COVID-19 pandemic, buye s o package a els showed loyal y o a el agencies. The use o machine lea ning allowed us o iden i y he easons o choosing a a el agency, which ha e a key impac on he loyal y o holiday package buye s du ing he pandemic. These ac o s belong p ima ily o he ca ego y o loyal y mo i es in he cus ome -p o ide ela- ionship ( us , indi idual app oach o o ice s a ) and he ca ego y o mo i es ela ed o he e al- ua ion o bene i s (belie in he alue o he o - e , a wide selec ion o ou is packages and e icien /ad iso y se ice). These i e mo i es eme ged as he mos impo an ac o s o each ype o loyal y (a ec i e, beha iou al and o al). The only di e ence was in he o de o hei occu ence. Buye s’ belie in he alue o he o - e , us , indi idual app oach o o ice s a , e icien /ad iso y se ice and a wide selec ion o ou is packages de e mined cus ome s’ p omo ion o posi i e opinions abou a gi en a el agency. In u n, he belie in he alue o he a el agency’s o e , us in he a el agency, an indi idual app oach o he a el agency’s employees, e icien /ad iso y se ice in he a el agency, he epu a ion and p es ige o he a el agency, as well as a wide selec- ion o ou is packages, in luenced he epea pu chase o package a els in he gi en o ice. Howe e , he shaping o he loyal y o pu chas- e s o package a els in a global pe spec i e, combining bo h he a ec i e and beha io al unde s anding o loyal y, was in luenced mos by belie in he alue o he o e , us , indi idual app oach o a el agency employees, e icien se ice in he o ice ha is able o ad ise a el and a wide selec ion o ou is packages. The i s posi ion in he anking o impo ance was always occupied by belie in he alue o he o e . I should be no ed ha his is a e y b oad and agg ega e concep , and o he ac o s may also wo k o in luence he deg ee o buye belie in he alue o an o e . The s udy p esen ed he e en iches knowl- edge in he a ea o ou is loyal y managemen and ou is beha iou . The esea ch ecognises he an eceden s o loyal y in an unp eceden ed si ua ion – a pandemic. I con ibu es o mul i- plying he knowledge o he impo ance o in- di idual loyal y mo i a o s and hei ele ance depending on he si ua ion o he en i onmen . I p o es he cons an ele ance o he issue o pu chasing beha io and he ac o s de- e mining i , as well as he need o epea ed esea ch in his a ea, especially in he changing condi ions o ope a ion o ou is en e p ises. The o iginali y o he esea ch lies in he use o a new me hod o da a analysis, i.e., machine lea ning. Al hough he esul s achie ed a e no su p ising and he iden i ied loyal y mo i es do no di e om hose ound ou by p e ious esea che s be o e he COVID-19, he con i - ma ion o he indings indica es ha he new me hod is aluable and use ul. Mo eo e , i s addi ional ad an age is ha he esul s ob ained ha e a p edic i e alue. Wi h a high deg ee o p obabili y he iden i ied se o ac o s will in- luence he loyal y beha io o ou ism package buye s in he u u e. This is o g ea impo ance o he u ili y o he esea ch esul s in p ac- ice. This a ibu e can con ibu e o imp o ing he e ec i eness o u u e ma ke ing decisions o companies, including a el agencies. Ma ke ing and T ade 217 2024, olume 27, issue 2, pp. 201–220, DOI: 10.15240/ ul/001/2024-2-012 The esul s o his s udy may be use ul o package a els p o ide s as a sou ce o in- o ma ion o planning mo e e ec i e ma ke - ing s a egies and c isis eco e y s a egies. The indings sugges ocusing on he ac i i ies s eng hening cus ome us in he a el agency, placing an emphasis on imp o ing he company’s image and p epa ing app op ia e s anda ds o main ain he e iciency o se ice in he e en o dis up ions in he se ice p o i- sion p ocess. Bo h he iden i ica ion o ac o s and a ho ough unde s anding o hei impac on he de elopmen o buye loyal y a e impo - an s eps om he poin o iew o he possi- bili y o e ec i e implemen a ion o ma ke ing ac i i ies aimed a building and s eng hening cus ome loyal y. As us and belie s a e buil o e he long e m, he ac i i ies o a el agen s should be o ien ed especially owa ds o ming posi i e, s ong and las ing ela ionships wi h cus om- e s. Rela ionship ma ke ing ools will be use ul he e. The cus ome decides on a long- e m ela ionship wi h a se ice p o ide when he o she pe cei es he ela ionship as bene icial. In he case o a el agencies, he cus ome bene i s on he one hand om a aluable o e o a el packages and on he o he om good se ice in he o ice. The o e o a a el agency is he esul o he agency’s coope a ion wi h ou ope a o s – supplie s o a el packages. Cus ome sa is ac ion wi h ega d o he alue o he o e depends on he abili y o a el agencies o a ac well-known a el b ands, bo h in e na ional (e.g., TUI, Necke mann) and domes ic ones (e.g., I aka, Rainbow, G ecos), o coope a e. T a el agen s should conside he wides possible coope a ion wi h mul iple ou ope a o b ands due o he ac ha cus ome s, acco ding o he esea ch, a e in e es ed in a wide choice o a el packages. The sa is ac ion o a el agency cus ome s also depends on he quali ica ions o employees, hei compe ences and a i ude o wo k. The e- sponden s hemsel es emphasised he de- pendence o epea ing a pu chase in a a el agency and ecommending he agency o o he cus ome s on he e icien ad iso y se ices and an indi idual app oach om he agency’s s a . A well-p epa ed s a ec ui men p o- cess, cons an imp o emen o employees, and equipping hem wi h he igh wo king ools a e he ounda ions o ensu ing a high le el o se ice. As e e y s udy, also his s udy is no wi h- ou limi a ions. Fi s ly, he limi a ions conce n he loyal y es ing me hod i sel . The da a is decla a i e and e e s o he u u e (in en ions). Secondly, compa isons o he esul s wi h he indings o p e ious esea che s mus be ca e ul because hey will co e di e en g oups o esponden s, and di e en se s o loyal y ac o s, and he da a was analysed using di - e en me hods. Thi dly, he esea ch only co e s a sample d awn om he Polish popula- ion, which makes i impossible o gene alise he esea ch esul s. In he u u e, aking in o accoun he dy- namics o he en i onmen , i is necessa y o con inue and epea he s udy o p edic o s o loyal y among buye s o ou is se ices in sea ch o changes ha may be impo an o he e ec i eness o aking ac ions o acqui e and e ain cus ome s. The esul s o simila s udies conduc ed in di e en cul u al a eas (coun ies), as well as add essing he issue o cus ome loyal y in online e sus o line a el agencies, may be o in e es . Acknowledgmen s: The esea ch was unded by he Minis y o Science and Highe Educa ion in Poland unde he p og amme “Regional Ini ia i e o Excellence” 2019–2022, p ojec numbe 015/RID/2018/19, o al unding amoun 10,721,040.00 PLN.” Re e ences Abou-Shouk, M. A., & Khali a, G. S. (2017). The in luence o websi e quali y dimensions on e-pu chasing beha iou and e-loyal y: A com- pa a i e s udy o Egyp ian a el agen s and ho els. Jou nal o T a el & Tou ism Ma ke ing, 34(5), 608–623. h ps://doi.o g/10.1080/105484 08.2016.1209151 Akba i, M., Meh ali, M., SeyyedAmi i, N., Rezaei, N., & Pou jam, A. (2020). Co po a e social esponsibili y, cus ome loyal y and b and posi ioning. Social Responsibili y Jou - nal, 16(5), 671–689. h ps://doi.o g/10.1108/ s j-01-2019-0008 Albay ak, T., Ka asakal, S., Kocabulu , Ö., & Du sun, A. (2020). Cus ome loyal y owa ds a el agency websi es: The ole o us and he- donic alue. Jou nal o Quali y Assu ance in Hos- pi ali y & Tou ism, 21(1), 50–77. h ps://doi.o g/ 10.1080/1528008x.2019.1619497 Ama o, S., & Dua e, P. (2015). An in eg a- i e model o consume s’ in en ions o pu chase Ma ke ing and T ade 218 2024, olume 27, issue 2, pp. 201–220, DOI: 10.15240/ ul/001/2024-2-012 a el online. Tou ism Managemen , 46, 64–79. h ps://doi.o g/10.1016/j. ou man.2014.06.006 Ande son, R. E., & S ini asan, S. S. (2003). E-sa is ac ion and e-loyal y: A con ingency amewo k. Psychology & Ma ke ing, 20(2), 123–138. h ps://doi.o g/10.1002/ma .10063 And eassen, W. T., & Lindes ad, B. (1998). Cus ome loyal y and complex se ices: The im- pac o co po a e image on quali y, cus ome sa - is ac ion and loyal y o cus ome s wi h a ying deg ees o se ice expe ise. In e na ional Jou - nal o Se ice Indus y Managemen , 9(1), 7–23. h ps://doi.o g/10.1108/09564239810199923 A siz, O. Ako a, O., & Ce in, G. (2021). Ex- amining package ou is s’ expe ience on o e all package ou sa is ac ion and beha io al in en- ions: Fi s - ime e sus epea package ou is s. Anais B asilei os De Es udos Tu is icos-Abe , 11, 1–18. Bendapudi, N., & Be y, L. L. (1997). Cus ome s’ mo i a ions o main aining ela- ionships wi h se ice p o ide s. Jou nal o Re ailing, 73(1), 15–37. h ps://doi.o g/10.1016/ s0022-4359(97)90013-0 B eiman, L. (2001). Random o es s. Machine Lea ning, 45(1), 5–32. h ps://doi.o g/10.1023/ a:1010933404324 Campo, S., & Yagüe, M. J. (2008). Tou is loyal y o ou ope a o : E ec s o p ice p o- mo ions and ou is e o . Jou nal o T a el Resea ch, 46(3), 318-326. h ps://doi.o g/ 10.1177/0047287507303975 Ca uana, A. (2002). Se ice loyal y: The e - ec s o se ice quali y and he media ing ole o cus ome sa is ac ion. Eu opean Jou nal o Ma ke ing, 36(7/8), 811–828. h ps://doi.o g/ 10.1108/03090560210430818 Ce in, G. (2020). Expe ience s quali y: P edic ing sa is ac ion and loyal y in se ices. The Se ice Indus ies Jou nal, 40(15–16), 1167–1182. h ps://doi.o g/10.1080/02642069. 2020.1807005 Chaudhu i, A., & Holb ook, M. B. (2001). The chain o e ec s om b and us and b and a ec o b and pe o mance: The ole o b and loyal y. Jou nal o Ma ke ing, 65(2), 81–93. h ps://doi.o g/10.1509/jmkg.65.2.81.18255 Chen, C.-F., & Tsai, M.-H. (2008). Pe cei ed alue, sa is ac ion, and loyal y o TV a el p oduc shopping: In ol emen as a mode a- o . Tou ism Managemen , 29(6), 1166–1171. h ps://doi.o g/10.1016/j. ou man.2008.02.019 Conze, O., Biege , T., Laesse , C., & Riklin, T. (2010). Rela ionship in en ion as a media o be ween ela ional bene i s and cus- ome loyal y in he ou ope a o indus y. Jou - nal o T a el & Tou ism Ma ke ing, 27(1), 51–62. h ps://doi.o g/10.1080/10548400903534899 Day, G. S. (1969). A wo-dimensional con- cep o b and loyal y. Jou nal o Ad e ising Resea ch, 9(3), 29–35. Del Bosque, I. A. R., San Ma ín, H., & Col- lado, J. (2006). The ole o expec a ions in he consume sa is ac ion o ma ion p ocess: Empi ical e idence in he a el agency sec o . Tou ism Managemen , 27(3), 410–419. h ps:// doi.o g/10.1016/j. ou man.2004.10.006 De ece, C., Ga cia-Ag eda, S., & Ribei o- Na a e e, B. (2015). The alue o us o a el agencies in achie ing cus ome s’ a i udi- nal loyal y. Jou nal o P omo ion Managemen , 21(4), 516–529. h ps://doi.o g/10.1080/104964 91.2015.1051409 Dick, A., & Baśu, K. (1994). Cus ome loyal y: Towa ds an in eg a ed amewo k. Jou nal o he Academy o Ma ke ing Science, 22(2), 99–113. h ps://doi.o g/10.1177/0092070394222001 Dudek, A., Ja emen, D. E., Michalska- Dudek, I., & Walesiak, M. (2019). Loyal y model p oposal o a el agency cus ome s. Sus ain- abili y, 11(13), 3702. h ps://doi.o g/10.3390/ su11133702 Elga hy, S. D. (2022). E ec s o se ice quali y, loyal y p og ams, p icing s a egies, and cus ome engagemen on i ms’ pe o mance in Egyp ian a el agencies: Media ing e ec s o cus ome e en ion. Jou nal o Quali y Assu - ance in Hospi ali y & Tou ism, 24(6), 753–781. h ps://doi.o g/10.1080/1528008x.2022. 2077889 Fishe , A., Rudin, C., & Dominici, F. (2019). All models a e w ong, bu many a e use ul: Lea ning a a iable’s impo ance by s udying an en i e class o p edic ion models simul ane- ously. Jou nal o Machine Lea ning Resea ch, 20(177), 1-81. h ps://doi.o g/10.48550/ARXIV. 1801.01489 Goldbe g, S. M. (1981). An empi ical s udy o li es yles co ela es o b and loyal y beha iou . Ad ances in Consume Resea ch, 9, 456–460. Gonçal es, H. M., & Sampaio, P. (2012). The cus ome sa is ac ion-cus ome loyal y ela ionship: Reassessing cus ome and ela- ional cha ac e is ics mode a ing e ec s. Man- agemen Decision, 50(9), 1509–1526. h ps:// doi.o g/10.1108/00251741211266660 Gouna is, S. P. (2005). T us and commi - men in luences on cus ome e en ion: Insigh s Ma ke ing and T ade 219 2024, olume 27, issue 2, pp. 201–220, DOI: 10.15240/ ul/001/2024-2-012 om business- o-business se ices. Jou nal o Business Resea ch, 58(2), 126–140. h ps:// doi.o g/10.1016/s0148-2963(03)00122-x G emle , D. D., & B own, S. W. (1996). Se ice loyal y: I s na u e, impo ance, and implica ions. In B. Ed a dsson, S. W. B own, & R. E. Johns on (Eds.), Ad ancing se ice quali y: A global pe spec i e (pp. 171–180). In e na ional Se ice Quali y Associa ion, Inc., Business Resea ch Ins i u e. Han, H., Kim, Y., & Kim, E.-K. (2011). Cogni- i e, a ec i e, cona i e, and ac ion loyal y: Tes - ing he impac o ine ia. In e na ional Jou nal o Hospi ali y Managemen , 30(4), 1008–1019. h ps://doi.o g/10.1016/j.ijhm.2011.03.006 Huang, L. (2008). Explo ing he de e - minan s o e-loyal y among a el agencies. The Se ice Indus ies Jou nal, 28(2), 239–254. h ps://doi.o g/10.1080/02642060701842316 Jacoby, J., & Ches nu , R. W. (1978). B and loyal y: Measu emen and managemen . Wiley. Jas o ia, S. S., Mulchandani, K., & S i as- a a, S. (2023). D i e s o expe ien ial loyal y in en ions owa ds online a el agen s: A com- pa a i e analysis on p e- and pos -COVID a elle s. Consume Beha io in Tou ism and Hospi ali y, 18(4), 563–586. h ps://doi. o g/10.1108/cb h-12-2022-0200 Konecnik, M., & Ga ne , W. C. (2007). Cus ome -based b and equi y o a des ina ion. Annals o Tou ism Resea ch, 34(2), 400–421. h ps://doi.o g/10.1016/j.annals.2006.10.005 Lai, I. K. W. (2014). The ole o se ice qual- i y, pe cei ed alue, and ela ionship quali y in enhancing cus ome loyal y in he a el agency sec o . Jou nal o T a el & Tou ism Ma ke ing, 31(3), 417–442. h ps://doi.o g/10.1080/105484 08.2014.883346 Lawu, M. S., San oso, I., Asbudiu omo, R. M., & Willy Gunadi, W. (2022). Se ice eco e y in online a el agencies du ing COVID-19 pandemic. In P oceedings o he In e na ional Con e ence on Indus ial Engi- nee ing and Ope a ions Managemen Is anbul (pp. 4386–4397). IEOM. Lee, M., & Cunningham, L. F. (2001). A cos / bene i app oach o unde s anding se ice loyal- y. Jou nal o Se ices Ma ke ing, 15(2), 113–130. h ps://doi.o g/10.1108/08876040110387917 Lee, S., Jeon, S., & Kim, D. (2011). The im- pac o ou quali y and ou is sa is ac ion on ou is loyal y: The case o Chinese ou is s in Ko ea. Tou ism Managemen , 32(5), 1115–1124. h ps://doi.o g/10.1016/j. ou man.2010.09.016 Majo , B., & McLeay, F. (2013). Ali e and kick- ing: E alua ing he o e seas package holiday expe ience o g ey consume s in he Uni ed King- dom. Jou nal o Vaca ion Ma ke ing, 19(1), 5–18. h ps://doi.o g/10.1177/1356766712461405 Mazu ek-Kusiak, A. (2020). De e minan s o he selec ion o a el agencies on Polish ou is se ices ma ke . E&M Economics and Manage- men , 23(1), 156–166. h ps://doi.o g/10.15240/ ul/001/2020-1-011 Moisescu, O. I., & Gica, O. A. (2014). The impac o se ice quali y on cus ome be- ha io al loyal y in he case o a el agencies om Romania. Am i ea u Economic Jou nal, 16(8), 1191–1204. Mo gan, R. P. (1999). A consume -o ien ed amewo k o b and equi y and loyal y. In e na- ional Jou nal o Ma ke Resea ch, 42(1), 1–9. h ps://doi.o g/10.1177/147078530004200105 Mu sid, A., & Wu, C. H.-J. (2022). Cus ome pa icipa ion, alue co-c ea ion and cus ome loyal y: E idence om Um ah a el agen- cies in Indonesia. Jou nal o Islamic Ma ke - ing, 13(3), 628–648. h ps://doi.o g/10.1108/ jima-06-2020-0190 Nagelke ke, N. J. D. (1991). A no e on a gene al de ini ion o he coe icien o de e - mina ion. Biome ika, 78, 691–692. h ps://doi. o g/10.2307/2337038 Newman, J. W., & We bel, R. A. (1973). Mul i a ia e analysis o b and loyal y o majo household appliances. Jou nal o Ma ke ing Resea ch, 10(4), 404–409. h ps://doi.o g/ 10.1177/002224377301000408 Oli e , R. L. (1999). Whence consume loy- al y? Jou nal o Ma ke ing, 63(4_suppl1), 33–44. h ps://doi.o g/10.1177/00222429990634s105 P ause, G., Mendez, M. M., & Ga cia- Ag eda, S. (2013). A i udinal loyal y and us in en ep eneu ship: Building new ela ionships. In e na ional En ep eneu ship and Manage- men Jou nal, 9(4), 531–540. h ps://doi.o g/ 10.1007/s11365-011-0215-y Reichheld, F. R., & Sche e , P. (2000). E-loyal y: You sec e weapon on he web. Ha a d Business Re iew, 78(July–Augus ), 105–114. Richa d, J. E., & Zhang, A. (2012). Co - po a e image, loyal y, and commi men in he consume a el indus y. Jou nal o Ma ke ing Managemen , 28(5–6), 568–593. h ps://doi.o g/ 10.1080/0267257x.2010.549195 Rizal, H., Amin, H., Suddin, L., Son- doh, S. L., & Jung Ku, C. (2020). Rela ionship Ma ke ing and T ade 220 2024, olume 27, issue 2, pp. 201–220, DOI: 10.15240/ ul/001/2024-2-012 quali y and e-loyal y owa ds online a el agen- cy (OTA): Social exchange heo y pe spec i e. Ju nal Pengu usan, 58, 39–51. h ps://doi.o g/ 10.17576/pengu usan-2020-58-04 Roge -Monzó, V., Ma í-Sánchez, M., & Guija o-Ga cía, M. (2015). Using online con- sume loyal y o gain compe i i e ad an age in a el agencies. Jou nal o Business Resea ch, 68(7), 1638–1640. h ps://doi.o g/10.1016/ j.jbus es.2015.02.009 Ronsana, G. G., Shihab, M. R., Syahbud- din, B. H., & Fi iani, W. R. (2018). Fac o s in luencing cus ome ’s e-loyal y in ou ism e-ma ke place. In P oceedings o 2018 In- e na ional Con e ence on In o ma ion Tech- nology Sys ems and Inno a ion (ICITSI) (pp. 237–241). IEEE. h ps://doi.o g/10.1109/ ici si.2018.8696021 Rowley, J. (2005). The ou Cs o cus- ome loyal y. Ma ke ing In elligence & Plan- ning, 23(6), 574–581. h ps://doi.o g/10.1108/ 02634500510624138 Rundle-Thiele, S. (2006). Look a e me and I will look a e you! Jou nal o Consume Ma ke ing, 23(7), 414–420. h ps://doi.o g/ 10.1108/07363760610712957 San os, S., Mon ei o, J., Espí i o San- o, P. M., Augus o, L., & Oli ei a, A. (2022). The u u e o b and awa eness and b and loyal- y on a el agencies. In e na ional Con e ence on Tou ism Resea ch, 15(1), 371–378. h ps:// doi.o g/10.34190/ic .15.1.107 Sende s, A., Go e s, R., & Neu s, B. (2013). Social media a ec ing ou ope a o s’ cus ome loyal y. Jou nal o T a el & Tou ism Ma ke ing, 30(1–2), 41–57. h ps://doi.o g/10.1080/105484 08.2013.750993 Se ó-Pamies, D. (2012). Cus ome loyal y o se ice p o ide s: Examining he ole o se - ice quali y, cus ome sa is ac ion and us . To al Quali y Managemen & Business Excel- lence, 23(11–12), 1257–1271. h ps://doi.o g/ 10.1080/14783363.2012.669551 Sil a, G. M., & Gonçal es, H. M. (2016). Caus- al ecipes o cus ome loyal y o a el agencies: Di e ences be ween online and o line cus om- e s. Jou nal o Business Resea ch, 69(11), 5512–5518. h ps://doi.o g/10.1016/j.jbus es. 2016.04.163 S a is a. (2021). Online a el ma ke wo ld- wide. h ps://www.s a is a.com/s udy/15218/ on line- a el-ma ke -s a is a-dossie / Thi Phuong Thao, H., & Swie czek, F. W. (2008). In e ne use, cus ome ela ionships and loyal y in he Vie namese a el indus y. Asia Paci ic Jou nal o Ma ke ing and Logis- ics, 20(2), 190–210. h ps://doi.o g/10.1108/ 13555850810864551 Wasza u ys yka. (2020). Klienci, k ó ym poświęca się czas, lepiej eagują podczas k y- zysu [Cus ome s who a e gi en ime espond be e du ing a c isis]. h ps://www.wasza u- ys yka.pl/klienci-k o ym-poswieca-sie-czas- lepiej- eaguja-podczas-k yzysu/ Wasza u ys yka. (2021). Agen u ys yczny u p ogu sezonu le niego 2021 i w gąszczu co- idowych zasad [A a el agen a he h eshold o he summe season 2021 amids he maze o COVID egula ions] (Podcas No. 13). h ps:// www.wasza u ys yka.pl/agen - u ys yczny- u-p ogu-sezonu-le niego-2021-i-w-gaszczu- co idowych-zasad-podcas -13/) Wu, W., & Rian ama, D. (2022). De e min- ing he ac o s a ec ing cus ome sa is ac ion using an ex ac ion-based ea u e selec ion app oach. Pee J Compu e Science, 8, e850. h ps://doi.o g/10.7717/pee j-cs.850 Yang, A. J.-F., Chen, Y. J., & Huang, Y.-C. (2017). Enhancing cus ome loyal y in ou ism se ices: The ole o cus ome -company iden i- ica ion and cus ome pa icipa ion. Asia Paci ic Jou nal o Tou ism Resea ch, 22(7), 735–746. h ps://doi.o g/10.1080/10941665.2017. 1319398 Zhu, Z., Kim, S., & Shin, H. H. (2022). Does he websi e c edibili y o online a el agencies (OTAs) a ec cus ome s’ e alua ion and u u e in en ions? Asia Paci ic Jou nal o Tou ism Re- sea ch, 27(8), 842–855. h ps://doi.o g/10.1080 /10941665.2022.2131444