Fixing Instagram's Content Discovery Problem
Five interviews and sixteen survey responses on how Instagram users actually handle bad recommendations
✳The problem they went after
The question was how Instagram users experience content discovery, recommendation quality and the controls meant to personalise the feed. Recruiting covered five distinct behaviour segments: passive consumers, active posters, small business creators, DM-first users and exploration-driven browsers. What came back was not a missing control but an unused one, which is a much harder problem to fix.
✳What they actually did
Five in-depth interviews across those segments, sixteen survey responses for quantitative scale, and seven insights each carried by a direct user quote alongside the survey data. The headline is that when recommendations go wrong, users do not correct them, they scroll and hope. Seven of sixteen respondents defaulted to passive coping against six who used Not Interested. Two starting hypotheses were invalidated: Explore as the main discovery surface, since Reels accounted for nine of sixteen last-enjoyed discoveries, and users as self-correcting algorithm trainers.
✳What you can take from it
Worth stealing
- Five behaviour segments defined before recruiting, not after
- Every insight paired with a quote and survey data
- Two starting hypotheses reported as invalidated
- Passive coping reframed as a trust finding
Read this if
Running discovery research and wanting a model for reporting it honestly
The transferable bitA low usage number on a feedback control looks like a discoverability problem until you ask why, and here it turns out people stopped believing feedback changes anything. Three unprompted insights followed, including Instagram serving different jobs by user type and users leaving for Pinterest or WhatsApp when it fails them.
✳The submission itself
Open the PDF →Reading on a phone? The embedded viewer is cramped — open the PDF in a new tab instead.
