Bolta Research
Most numbers about AI content are guesses. These are not. Bolta Research publishes benchmarks measured from our own product data — what humans actually do with an AI-written draft, how much they edit it, whether it publishes, and how often teams post once they keep going. Free to read, free to cite, free to embed.
AI made social content cheap to produce. Approval did not scale with it. Across 104 workspaces and 11,258 AI-generated posts, 46.4% expired waiting for a decision and a further 16.7% was never submitted for one — while just 3.0% was ever reviewed and turned down on merit.
Every figure is aggregated across workspaces — never a single customer, never identifiable post content. A cell is only reported when it clears a stated minimum cohort size; below that threshold we publish nothing rather than publish noise. Each study names the exact date range it covers, what was excluded and why, and the limitations that would make us distrust our own numbers. If a required field is missing, the study stays unpublished and unlinked instead of shipping with a gap.