The Limits of Reputation in Platform Markets

on YouTube ↗50m

" summary

Uses eBay internal data on 935,000 new buyers tracked ~3 years (15M purchases, 1.8M sellers) to show that traditional percent-positive scores are biased by silent buyers and retaliation; defines effective percent positive (positives ÷ total transactions) and finds sellers in the top EPP quartile raise buyer return-to-eBay probability by about 4% and return-to-seller by ~2%. Natural-language signals from post-sale messages (the negative message rate) predict dissatisfaction as strongly as EPP. A randomized search demotion experiment (10% traffic, ~3pp EPP shift) changes outcomes without lowering short-run conversion.

AI-written summary · about

" chapters (14)

1. 0:00Intro
2. 0:42Platform Markets and Quality Control
3. 3:53Contributions
4. 10:12Conceptual Framework
5. 12:43What do buyers use to form expectations? Reputation! After every eBay transaction
6. 20:46Feedback is Biased
7. 23:19Effective Percent Positive (EPP) EPP
8. 24:27EPP Distribution
9. 27:37The Distribution of Buyer Purchases
10. 28:28The Scope for Externalities is real
11. 28:59The Main Regression: "Loyalty. (Voice) and Exit"
12. 31:29Main Regression Results: Ever Return
13. 38:51Measuring Treatment Effect: Discounted Search EPP
14. 43:31Concluding Remarks

" related talks

• 20mBoomerang: Rebounding the Consequences of Reputation Feedback on Crowdsourcing PlatformsSharon Zhou, Sachin Sharma, Dinesh Dhakal, Michael S. Bernstein, Kristy Milland, Rajan Vaish, Snehalkumar Neil S. Gaikwad, Durim Morina, Adam Ginzberg, Catherine Mullings, Shirish Goyal, Dilrukshi Gamage, Christopher Diemert, Mathias Burton, Mark Whiting, Karolina Ziulkoski, Alipta Ballav, Aaron Gilbee, Senadhipathige S. Niranga, Vibhor Sehgal, Jasmine Lin, Leonardy Kristianto, Angela Richmond-Fuller, Jeff Regino, Nalin Chhibber, Dinesh Majeti, Kamila Mananova, William Dai, Victoria Purynova, Samarth Sandeep, letshine Chandrakanthan, Tejas Sarma, Sekandar Matin, Ahmed Nasser, Rohit Nistala, Alexander Stolzoff, Vinayak Mathur · UIST 2016
• 33mPredictive NPS (PNPS)Martin Dimov · DSC Europe 2022
• 29mClearing Matching Markets Efficiently: Informative Signals and Match RecommendationsItai Ashlagi · Societal Networks 2018
• 52mDashboard Mechanisms for Online MarketplacesJason Hartline · Information Design and Data Science 2019
• 44mClearing Matching Markets Efficiently: Informative signals and Match RecommendationsPeng Shi · Platform Markets 2019
• 50mQuality Selection in Two-Sided Markets: A Price Discrimination ApproachRamesh Johari · Information Design and Data Science 2019
• 49mFrom research to production at ebayChristopher Severs, Alex Zhicharevich · Scala.IO 2014
• 50mDiversity in Matching MarketsJohn Dickerson · Information Design and Data Science 2019