Ranking by engagement and non‐engagement signals: Learnings from industry
Many online media platforms primarily rank items by predicted user engagement, yet we see substantial potential in incorporating non-engagement signals-such as user surveys or quality metrics-to benefit both platforms and society. Drawing on a daylong workshop with experts from various industry platforms and academia, we developed a series of propositions related to non-engagement signals that gained broad consensus. These propositions examine the relationships between engagement, non-engagement signals, and key metrics like user retention. We document each proposition with publicly available evidence, including quantitative results where possible. Given the cost and time associated with industry experimentation on non-engagement signals, we believe sharing these insights will be valuable for advancing common knowledge.
Authors
- Mohit Kothari (ORCID: https://orcid.org/0000-0002-8191-9769)
- Ravi Iyer
- Sanjay Kairam (ORCID: https://orcid.org/0000-0001-8320-222X)
- Smitha Milli (ORCID: https://orcid.org/0009-0000-0395-7025)
- Jonathan Stray (ORCID: https://orcid.org/0000-0003-4467-1239)
- Bonnie L. Barrilleaux
- Leif Sigerson
- Sana Pandey
- Behnam Rezaei (ORCID: https://orcid.org/0000-0001-9446-2473)
- J. Allen
- Tom Cunningham
Institutions
- University of Southern California (US)
- LinkedIn (United States) (US)
- Cornell University (US)
- Integrity Systems (United States) (US)
- Pollinator Partnership (US)
- OpenAI (United States) (US)
- University of California, Berkeley (US)
Publication Details
- Journal
- Annals of the New York Academy of Sciences
- Published
- 2025-08-20
- DOI
- https://doi.org/10.1111/nyas.15399
- Citations
- 1
- Primary Topic
- Digital Marketing and Social Media
- Type
- article
- Field-Weighted Citation Impact
- 4.20
Funders
- University of Southern California
- University of California Berkeley