Instagram Expands Algorithm Controls Test: Reach Math in Play

Instagram has begun testing expanded algorithm controls inside its main feed and Reels surfaces, extending a year-long push that started with the "Interested" toggle — a one-tap "I'm not interested" signal for Explore and Reels recommendations — rolled out in 2024.

The new test was surfaced by Social Media Today, which has a strong track record on early Meta (Instagram and Facebook's corporate parent) experiments, including the original "Interested" pilot. Meta hasn't published a help center entry or developer note on the rollout, and hasn't named the surfaces in scope.

What does the test actually change?

Specifics remain thin. Public reporting frames the experiment as a continuation of Instagram's control-testing program, suggesting the platform is moving from a single negative-feedback button toward a fuller toolkit closer to TikTok's layered "Not Interested" and topic-filter options (granular on/off switches for content categories) that have existed for years.

For practitioners, the practical question is whether the reach math shifts. It probably will, once the controls ship at scale.

How this hits creators and ad budgets

Algorithm controls reshape the recommendation graph — the network of "user-to-content" connections the ranking model uses to decide what appears next — in two ways social teams should plan around.

First, more negative-feedback options let creators lose reach faster when content misfires. A single "not interested" tap today only nudges the model; a sustained negative signal will re-weight future distribution more aggressively.

Second, audience composition tightens. Cleaner negative data means the platform knows more precisely who wants a niche topic, which compresses reach for broad-appeal posts and rewards topical depth.

For paid teams, the implication is familiar. Sharper organic signals typically push CPMs (cost per thousand impressions, the price advertisers pay per 1,000 feed views) upward on broad targeting, because the algorithm already knows who wants what. Niche interest targeting and creator-led whitelisting (running ads through a creator's handle to borrow their audience graph) should outperform lookalikes (audiences modeled to resemble an existing follower base) over the next two quarters if the controls roll out widely.

What's still unconfirmed

Treat feature names, rollout geography, and the specific surfaces in scope as provisional until Instagram confirms them in-app or via its @creators channel. Meta kills a meaningful share of these tests before they ship.

The watch item for the next 30 days: whether the test surfaces in more markets, and whether Meta ties the controls to its expanded "Reset Suggested Content" feature, which already lets users wipe their Reels history clean. If the two merge, creators shouldn't be surprised by a sharper, more volatile reach curve heading into Q1 — and they'd be wise to plan creative testing budgets accordingly.