Skip to main content
Free tool · Rate-limited

AI Overview likelihood predictor

Will Google's AI Overview trigger for your query? Claude predicts likelihood and the source types it'll cite.

Phrase it the way a real searcher would type into Google.

Affects local-pack triggers and country-specific source preferences.

Enter a query above. Free for 3 predictions per day per IP.

What this predictor actually estimates

Google's AI Overview (formerly Search Generative Experience) doesn't fire for every query. It triggers most often when the searcher is looking for a synthesized explanation, a comparison, or a multi-step answer — and rarely for navigational, transactional, or sensitive (medical / financial / legal) queries.

The predictor uses Claude to reason about your specific query against the documented and observed AIO trigger patterns. You get back a likelihood score, a verdict (very_likely / likely / possible / unlikely), the source types AIO is likely to pull from, and concrete content moves to make your page one of those sources.

How AI Overview decides what to cite

AI Overview pulls from sources it can quote without context. That means clean answer passages, well-structured listicles, expert-bylined explainers, and high-trust authority sites. It tends to skip jargon-heavy thought leadership, sales-y landing pages, and content that only makes sense if you've read the rest of the page.

Reading the “source types” the predictor returns is the most actionable signal. If AIO is going to cite Reddit threads and forum answers for your query, you can't compete with a corporate blog — you need contributor accounts on those forums plus your own page formatted like a Q&A. If AIO is going to cite official docs, your job is to be the official-looking explainer.

After running this — what to do next

  1. Run the AI Search visibility check on the same query to see who's actually getting cited right now in ChatGPT, Perplexity, Gemini, and Claude. AIO tends to mirror their picks closely.
  2. If you're not in the citation list, audit the cited pages — what format do they use? Long-form? Listicle? Q&A? Mirror that format on your own page.
  3. Apply the content tips above. The most common change is rewriting your intro to be a direct one-paragraph answer instead of a marketing pitch.
  4. Add structured data via our schema generator (FAQ or HowTo for explainer queries) and an llms.txt file so AI crawlers know which pages to favor.

FAQ

How accurate is the prediction?

It's directional. The predictor reasons from documented AIO trigger patterns + observed behavior, but Google's actual decision depends on signals only Google has (per-query model confidence, page-level quality signals, recency). Treat the verdict as 'should I optimize for AIO on this query?' rather than 'will it definitely show?'.

Why does the same query give different verdicts on different days?

AIO triggers shift as Google's confidence threshold shifts. A query at 65% likelihood today might be 50% next week. Use the predictor as a strategic input — if a query is consistently 'likely' or 'very_likely', it's worth investing content effort. Borderline 'possible' queries are riskier bets.

Does location really matter?

Sometimes a lot. AIO triggers more conservatively in regulated markets (medical, financial). For local-intent queries, the source mix shifts to Google Business Profile data. For comparison queries, US-vs-UK can pull entirely different cited sources. Setting location helps Claude weight those factors.

Why doesn't AIO trigger for my product/brand search?

Navigational and brand-only queries almost never trigger AIO. Google trusts that the user already knows the destination — there's no 'overview' to write. AIO opportunity is in the informational and commercial-investigation phases that happen before someone searches your brand.

How is this different from the AI Search visibility check?

This predicts whether AIO will fire and what kinds of sources it'll prefer (theory). The AI Search visibility check actually runs the query against an LLM with web search and shows who's currently cited (ground truth). Use them together: the predictor sets your strategy, the visibility check verifies it.

Will my query be saved?

No. We don't log inputs to this endpoint. Each run is processed by Claude and discarded.

Get the full Web Design Service workspace.

Want continuous AI Overview monitoring across every keyword? Sign up free →

No credit card required· 14-day trial