aeo/summary-bait fires when 70% or more of a page's citable facts are crammed into its first 150 words and nothing fresh waits below, a low-confidence warning that the page is shaped for an AI Overviews snippet Google can lift whole rather than for a reader who scrolls past the opener.
Optimize your content to trigger Google AI Overviews and answer engine summaries:
aeo/summary-bait measures one ratio: of all the citable facts on a page, what fraction sits in the first 150 words? The rule extracts facts with the same patterns aeo/citable-facts uses (dollar amounts, percentages, timeframes like '11 days' or '4 weeks', month-day dates, and form numbers) once across the whole page and once across the opener alone, then divides the opener count by the full count.
When 70% or more of the page's facts land in that opener, and the page has at least 3 facts to begin with, the rule warns at low confidence. Two gates keep it quiet on healthy pages. First, the opener must already pass aeo/answer-first (a complete, fact-bearing lead) because front-loading a clear answer is good, not a fault. Second, the page must carry no interactive, downloadable, or gated value below the fold: a foraging-calendar widget, a printable spore-print key, or a sign-in-to-continue block all mean there is a real reason to scroll, so the rule stays silent. Only the overlap (strong opener, everything cited up top, nothing new beneath) trips it.
The nuance is the whole point. A page that answers the question in its first paragraph is doing the right thing, aeo/answer-first rewards exactly that, and an AI engine will happily cite a clean opening line. The failure aeo/summary-bait catches is one step further: a page that dumps every number, date, and figure into the opener and then pads the rest with filler that adds nothing a reader could not get from the snippet alone.
That shape is optimised for the machine at the expense of the human. When 70% of your facts live in 150 words, an AI Overview can lift the whole answer and the click never happens, the searcher gets what they need from the summary and the scroll dies on the fold. The fix is not to weaken the opener but to give the body a reason to exist: distribute facts so the full picture requires reading on, and add value a summary cannot carry. A page that earns the scroll keeps the reader; a page that bait the summary trades a visitor for a citation.
/forage/morel-season, an urban-foraging field guide whose 150-word opener states everything: morels emerge when soil holds at 50 degrees for 4 weeks, the spring window runs roughly April 14 to May 26, a healthy patch yields 26% more by weight near dead elms, and a good spore print sets in 11 days. The 600 words beneath repeat the same claims in looser prose, add no new figure, and link to no tool. 4 of the page's 5 citable facts sit in the opener (80% concentration) so the rule warns: an AI Overview can quote the whole morel calendar without ever sending the forager to the page.
/forage/morel-season: the same field guide, rebalanced. The opener still answers cleanly (morels fruit when the soil hits 50 degrees), but the dated season table, the 26%-near-elms yield data, a spore-print method that sets in 11 days, and a printable hedgerow-by-hedgerow foraging-basket checklist now live in sections below the fold. Fewer than 70% of the facts sit up top, an interactive harvest-calendar widget gives a real reason to scroll, and the snippet can no longer carry the full answer; the reader has to land on the page to get the ramps and chanterelle windows too.
aeo/summary-bait is an answer-engine rule, not a spam classifier; it never escalates into the critical spam tier, because front-loading facts is a forecast about zero-click exposure, not evidence of manipulation. It measures page shape: a strong opener, every citable fact concentrated in the first 150 words, and no interactive or downloadable value waiting below. That overlap is the worst case for an AI Overview; the engine can answer the query from the summary alone and the click-through never arrives.
The rule sits beside aeo/answer-first deliberately, as its mirror. answer-first asks whether the opener resolves the question for a machine that may only read the top; summary-bait asks whether the page left anything for the human who keeps scrolling. The two are not in tension, a healthy page passes both, with a clean lead and a body that still rewards the scroll. The danger it flags is the page that wins the snippet and loses the reader, and on a foraging guide that means an AI Overview reciting your morel calendar while the forager never opens the page that knows where the chanterelles are.
Want to know whether this rule actually fires on your site?
Run pseolint against your sitemap. The audit is free, takes about a minute, and returns a per-URL list of every rule that fired (including this one) with the exact metric values so you can prioritise the fix queue.