The Dark Library Effect is the divergence between how much an AI engine reads a page and how little traffic it sends to that page. AI crawlers absorb the content, summarise it in an answer, and satisfy the user in place, so the page is heavily read yet rarely clicked. The term was coined by Orbit Media (Andy Crestodina, mid-2026) for the page type where the gap is widest: articles and blog posts.
In their analysis of 560,000+ AI crawl requests across 74 sites (Cloudflare AI Crawl Control data, single-vendor and directional), article pages received about 8.7 percentage points less referral share than their crawl share predicted, homepages about 10.4 points more, and 47% of all pages earned zero referral visits despite being crawled.
The effect does not mean article content is worthless. It gets cited and trains what the model knows about a brand. Value a click counter never captures. It is distinct from dark traffic (real visits misfiled as "direct"): the Dark Library Effect is about visits that never happen at all, a form of zero-click discovery. The lesson is to judge articles by citations, not clicks.