Read out

Online feedback: How the wrong system can distort the picture

Espe­cially on­line shop­pers often base their pur­chas­ing de­cisions on cus­tomer re­views. Hardly anyone books a hotel without first scrolling through all the feed­back pos­ted by pre­vi­ous guests. In his re­search, WU Pro­fessor Ben Greiner, head of the In­sti­tute for Mar­kets and Strategy and the Com­pet­ence Center for Ex­per­i­mental Re­search at WU, in­vestig­ates the factors that cause dis­tor­tions in on­line feed­back sys­tems. Such dis­tor­tions are det­ri­mental not only to cus­tomers, who are provided with less re­li­able in­form­a­tion, but also to busi­nesses, be­cause their cus­tomers are less likely to trust them due to less re­li­able in­form­a­tion.

Whether it’s book­ing a hotel or shop­ping on­line: Many people base their de­cisions on cus­tomer re­views. At the same time, pub­licly pos­ted re­views provide a strong in­cent­ive for com­pan­ies to im­prove products and ser­vices. For this to work, cus­tomer feed­back has to be both hon­est and in­form­at­ive. WU Pro­fessor Ben Greiner uses field data ana­lysis of on­line mar­ket­places like eBay and lab­or­at­ory tests to in­vestig­ate which on­line feed­back sys­tems work best. He ob­served a num­ber of dif­fer­ent pat­terns of so­cial be­ha­vior that were af­fected by the feed­back rules of each sys­tem.

Re­venge feed­back for neg­at­ive re­views

On eBay, one of the mar­ket plat­forms that al­lows buy­ers and seller to provide mu­tual feed­back, Greiner’s study provides evid­ence for strong re­cipro­city in feed­back be­ha­vior. Pos­it­ive re­views from buy­ers were al­most al­ways re­war­ded with pos­it­ive feed­back from vendors, whereas a neg­at­ive re­view al­most al­ways res­ul­ted in neg­at­ive re­venge feed­back from the seller. As a res­ult, many people who had a neg­at­ive ex­per­i­ence with a pur­chase chose not to leave feed­back at all, for fear of get­ting neg­at­ive feed­back in re­turn. This dis­torts the entire feed­back im­age: 98% of the feed­back given on eBay is pos­it­ive. “This means that the col­lect­ive feed­back of­fers only a lim­ited amount of in­form­a­tion,” says Greiner, “Our lab­or­at­ory stud­ies have shown that this is the res­ult of the sys­tem’s open­ness. If you change the rules and don’t pub­lish feed­back until both parties to the trans­ac­tion have sub­mit­ted their re­view, it leads to more neg­at­ive, hon­est feed­back but, un­for­tu­nately, also to a lower rate of par­ti­cip­a­tion in the feed­back sys­tem as a whole. It also works bet­ter if only one-sided feed­back is al­lowed.”

Poor com­prom­ise

Al­low­ing posters the op­tion to re­tract feed­back may also have un­in­ten­ded ef­fects. Ideally, this op­tion al­lows both parties to the trans­ac­tion to find a solu­tion in the event of a con­flict. The stud­ies re­vealed the op­pos­ite be­ha­vior, however: For example, sellers who were ex­pect­ing neg­at­ive re­views pos­ted neg­at­ive feed­back about buy­ers as lever­age to later con­vince the buy­ers to re­tract their own neg­at­ive re­view. “This also res­ults in feed­back dis­tor­tion. The in­form­a­tion value and ef­fect­ive­ness of the feed­back sys­tem are con­sid­er­ably re­duced,” says Greiner. Over­all it was shown that the spe­cific rules of in­di­vidual feed­back sys­tems de­termine the stra­tegic in­cent­ives buy­ers and sellers are ex­posed to when they con­trib­ute their feed­back. An ef­fi­ciently designed and planned feed­back sys­tem takes these factors into ac­count.

The study

For his re­search, Ben Greiner used field data from eBay and other on­line mar­ket plat­forms as well as data from lab­or­at­ory ex­per­i­ments. Greiner ex­plains, “In the field, there are a num­ber of vari­ables we can’t ob­serve. For this reason, we re­cre­ate mar­kets in the com­puter lab­or­at­ory and ask hu­man par­ti­cipants to make de­cisions tak­ing the roles of buy­ers and sellers. The profits that par­ti­cipants made in the lab ex­per­i­ments were paid out to them in cash. This en­sures the au­thenti­city of the de­cisions made. Like in a wind tun­nel, we can then change the design, i.e. the rules of the feed­back sys­tem, and ob­serve the re­ac­tions of the buy­ers and sellers.”

Ben Greiner

Ben Greiner has been head of WU’s In­sti­tute for Mar­kets & Strategy and the Com­pet­ence Center for Ex­per­i­mental Re­search, in­clud­ing the WULabs, since 2016. Ori­gin­ally from Ger­many, he stud­ied busi­ness ad­min­is­tra­tion at Hum­boldt-Uni­versität zu Ber­lin, earned his doc­tor­ate at the Uni­versity of Co­logne, and then spent two years as a re­searcher at Har­vard Uni­versity. Be­fore com­ing to WU, Greiner taught at the Uni­versity of New South Wales in Sydney, Aus­tralia. Ben Greiner’s re­search fo­cuses on fun­da­mental ques­tions of eco­nomic in­ter­ac­tion in so­cial con­texts. This in­cludes, for example, how groups gen­er­ate joint de­cisions based on the opin­ions of the group mem­bers, or how un­cer­tainty about the pref­er­ences and strategies of other act­ing agents af­fects our co­oper­at­ive and mar­ket de­cisions. He also in­vestig­ates more ap­plied re­search ques­tions like design­ing and im­prov­ing real mar­kets, in­sti­tu­tions, and or­gan­iz­a­tions. His re­search con­trib­uted to mar­ket plat­form eBay’s de­cision to re­con­fig­ure their world­wide feed­back sys­tem and to the Aus­tralian gov­ern­ment’s plans to develop a CO2 emis­sions trad­ing sys­tem. His work has been pub­lished in top journ­als like the Amer­ican Eco­nomic Re­view, the Eco­nomic Journal, the Journal of Pub­lic Eco­nom­ics, Man­age­ment Science, and oth­ers.

En­gin­eer­ing Trust: Re­cipro­city in the Pro­duc­tion of Repu­ta­tion In­form­a­tion.
Man­age­ment Science 2013, ht­tps://

Dis­pute Res­ol­u­tion or Es­cal­a­tion? The Stra­tegic Gam­ing of Feed­back With­drawal Op­tions in On­line Mar­kets. Man­age­ment Science 2018, ht­tps://

Anna Maria Schwendinger
PR Man­ager
Tel: + 43-1-31336 ext. 5478
Email: anna.schwendinger­

Back to overview