A new study from the Institute for Public Policy Research found that only 18% of the top posts in a typical British social media feed come from people the user actually knows (IPPR, April 2026). The remaining 82% is split between influencers, recommended content, and ads.
The viral Hacker News thread on Sunday framed it bluntly: it is fads, not friends, which now dominate social media. The numbers behind the framing make the case stronger than the headline.
What IPPR actually measured
IPPR commissioned the polling firm Survation to survey 1,000 UK adults in February 2026. Participants were asked to open their most-used social platforms — Instagram, Facebook, X, Bluesky, and TikTok — and categorize the top four posts that appeared.
The aggregated result:
| Category | % of top posts |
|---|---|
| People users know (friends, family) | 18% |
| Influencers, public figures, recommended content | 35% |
| Advertisements and branded material | 29% |
| Other | 18% |
Why it matters: social media platforms are funded by the same incentive — keep users scrolling. The cheapest way to fill a feed with high-engagement content is not by showing people their friends. It is by showing them whatever the algorithm thinks they will react to. The 82% of feed that is not from someone you know is the price of that incentive.
Platform breakdown
Not every platform performs the same way. The cleanest internal contrast in the IPPR data:
- TikTok and X show the lowest friend/family content — roughly 10% of top posts.
- Approximately 14% of posts on these platforms are advertisements.
- Instagram and Facebook show meaningfully more friend/family content than TikTok and X, though still under a third on average.
- Bluesky — the newest entrant in the study — sits between the two camps, reflecting its smaller advertising base but algorithmic For You feed.
The headline that travels — “fads, not friends” — is most defensible on TikTok and X. On Facebook, the story is more nuanced: your feed is dominated by groups and brands, not exactly strangers, but rarely your actual friends.
The political-isolation finding
One of IPPR’s sharper findings is buried in the demographic breakdown. Reform UK voters see known contacts in just 13% of their top posts — the most algorithmically isolated cohort in the study. Green party voters see known contacts at 23%, the least isolated.
Dr. Sofia Ropek-Hewson, the IPPR researcher who led the analysis, summarized it like this: platforms claiming to connect us are “making us less visible to each other and more isolated” through profit-driven algorithmic design.
That is the policy framing. It is also the marketing-language framing. The political-isolation gap is a useful argument for regulators because it offers a concrete harm — algorithmic sorting reinforces filter bubbles — without relying on the older “echo chamber” rhetoric that platforms have learned to deflect.
What IPPR is asking governments to do
The report makes three policy recommendations. Two are old. One is new.
- Amend the Online Safety Act to regulate manipulative algorithmic design. (Old — the UK’s Online Safety Act 2023 gave Ofcom partial authority here; IPPR wants it strengthened.)
- Require platforms to promote public-interest content from charities and community organizations. (Old — variants of this exist in the EU’s Digital Services Act.)
- Build a public-service social media platform led by the BBC and European broadcasters. (New — and the recommendation that will get all the headlines.)
The public-service social platform idea is not crazy on paper. The BBC and ARD both have decades of track record running large information platforms under public-service charters. The execution problem is that none of them have shipped a consumer-grade social product, and the network effects working against any new entrant in 2026 are nearly unbeatable.
Why the contrarian view is gaining traction
The IPPR study landed in a moment that was already shifting. Three things have happened in parallel:
- A “chronically offline” trend on TikTok itself — videos of users vowing to delete the app and replace it with vinyl records, brick phones, and in-person hobbies (CNBC, February 2026).
- A rise in paid in-person social events — speed dating, networking nights, and curated dinners — among users in their 20s who cite loneliness as a driver.
- Bluesky’s growth slowing — the most plausible “different feed” alternative is no longer growing fast enough to threaten the incumbents.
What is happening on the platform side is a quiet acknowledgement. Instagram has been re-promoting Stories from friends. TikTok added a “Friends” tab. X is testing a chronological feed for verified followers. None of those are policy changes. All of them are PR pre-emption.
What this means for builders and operators
If you work in social product or media policy, three takeaways:
- The friends-vs-feed split is now a measurable KPI. Expect platforms to start disclosing their own version of IPPR’s metric — likely after a regulator forces them to. The disclosure itself will move advertiser sentiment.
- Public-interest content slots are the next ad inventory. Whether or not the BBC-led platform ships, the policy direction is toward mandated non-commercial slots. Build for them.
- Smaller, friend-only platforms have a window. Geneva and similar group-chat-first products are seeing the kind of growth that early Discord saw. They will not replace Instagram, but they will eat the “what are my actual friends up to” job.
For more context, see our coverage of the EU Digital Services Act’s algorithmic transparency requirements and our analysis of the BBC’s 2026 digital strategy.
What to watch next
- Ofcom’s response to the IPPR report. Whether the regulator opens a formal investigation into algorithmic design under the existing Online Safety Act will signal the regulatory direction.
- EU AI Act enforcement on recommender systems. Algorithmic feeds are likely a high-risk classification — guidance is due in Q3 2026.
- Whether US regulators pick up the IPPR framing. “Friends vs strangers” is more defensible in Congress than “filter bubble.” Watch the FTC’s 2026 platform-competition report.
- A US replication of the IPPR survey. If Pew or a major academic group runs the same methodology in the US and gets similar numbers, the policy conversation widens dramatically.
The IPPR study did not invent the complaint. It put a defensible number on something that has been an intuition for a decade. Eighteen percent is a measurable harm. The question for the next twelve months is whether regulators have the appetite — and the legal toolkit — to move on it.
FAQ
What is the IPPR study?
The Institute for Public Policy Research, a UK think tank, commissioned Survation to survey 1,000 UK adults in February 2026 about what appears in their social media feeds. The report was published April 13, 2026.
What did the study find?
Only 18% of top posts on participants’ most-used social platforms came from people they actually knew. 35% came from influencers and recommended content, 29% from ads, and the rest from other categories.
Which platforms showed the worst friend-to-feed ratio?
TikTok and X both showed roughly 10% friend/family content in the top posts — the lowest in the study. Approximately 14% of posts on those platforms were advertisements.
Why does this matter for policy?
The numbers give regulators a measurable harm to point at. Vague “filter bubble” complaints have been easy for platforms to deflect; “your feed is 82% strangers” is harder to argue with.
What does IPPR want governments to do?
Three things: amend the UK Online Safety Act to regulate manipulative algorithmic design, require platforms to promote public-interest content, and build a public-service social platform led by the BBC and European broadcasters.
Is the BBC-led platform idea realistic?
Public broadcasters have decades of experience running large information platforms but none have shipped a consumer-grade social product. Network effects against any new entrant in 2026 are formidable. The idea is more plausible as policy leverage than as an actual product roadmap.
Why are Reform UK voters more algorithmically isolated?
The IPPR data shows Reform voters see known contacts in 13% of top posts, vs 23% for Green voters. The likely explanation: heavier consumption of recommended political content displaces friend posts in the ranking. The mechanism is algorithmic, not partisan.
Is anyone trying to fix this on the platform side?
Yes, but quietly. Instagram has been re-promoting friend Stories. TikTok added a Friends tab. X is testing chronological feeds for verified followers. None are policy concessions; all are PR pre-emption.
Track tech-policy developments as they land — subscribe to the TechDaily360 weekly briefing.
