Original research
Data-driven studies from the SocialToolsAI dataset. We run the analysis on real, public Instagram accounts and publish the numbers behind the takes that everyone has an opinion on.
Sports, history and UGC accounts gained followers between spring and October 2026. Crypto and meme pages lost them, and the niche with the best engagement shrank. Posting recently mattered more than the niche: accounts quiet for a year didn't grow, whatever they posted about.
About half of business accounts land a post at 5x their usual likes within six weeks, and most land more than one. The lone lucky tile is the rare case. We can't say why a post worked, but we can say how often it happens and what it doesn't tell you.
Globally, the best-vs-worst UTC hour gap is just 4.6% after within-account controls. But split by audience country, the effect is real and large: 57.7% for US/CA audiences, 43.9% for India/ID. The reason every 'post at Wednesday 9am' chart disagrees with every other one is that they all mix timezones into noise. The signal is only visible at the niche level.
Reels and photos engage almost identically on the median post - but 70% of the top 1% of posts by engagement rate, and 84% of the top 0.1%, are Reels. Same engagement floor. Much fatter right tail. The standard 'Reels-first' advice is half right, for the wrong reason.
Every guide tells you to slow down because per-post engagement drops. They're missing half the picture. We analyzed cadence and engagement data from 2,400 public Instagram accounts and found that total weekly engagement keeps climbing as posting frequency goes up - even when per-post rate falls.
More studies are in the works. Topics we're sitting on data for: niche-specific cadence curves, the half-life of a viral post, carousels versus single photos once that signal is clean, and what audience overlap actually predicts. Sign up to get them when they go live.
Every study on this page runs on the same underlying dataset that powers SocialToolsAI's Competitor Map and Rival Scout: public Instagram accounts found by the Competitor Map, with recent-post engagement recorded when each report ran.
We use medians instead of means wherever possible because a single viral post can distort a mean for weeks. We control for follower size by binning into cohorts (1k-10k, 10k-100k, 100k-1M, 1M+) because the same posting cadence behaves very differently at different audience scales.
Where a finding only holds in one cohort, we say so. Where the sample is too thin to draw a conclusion, we leave the conclusion out. The goal is studies that hold up - not headlines that don't survive a re-read.
A Rival Scout report finds the accounts actually in your niche from your audience, then shows what's working for them and what to post next. Launch price: $24.99/month until October 31, then $34.99.
Get your first report