Skip to content
Writing
the answer box

Not every question gets an answer.

The set is unstable and nobody outside the engines knows the rule. What you can know is which of your own questions trigger one today, which is the only version of this that is actionable.

Published 2026-08-20 | 4 minute read

Composed summaries do not appear for everything. They appear for questions a paragraph can usefully finish, and the boundary moves — the same query can trigger one this month and not the next.

The pattern, as far as it holds

  • Definitional and how-to questions: almost always.
  • Comparative questions with a small, stable option set: often.
  • Local questions with a geography: frequently, and the answer varies hard by where the request appears to come from.
  • Navigational and branded queries: rarely. Somebody typing your name wants you, and a summary adds nothing.
  • Anything ambiguous, contested or fast-moving: inconsistently, which is the honest description.

Treat all of that as a prior, not a rule. The engines change the threshold without announcing it, and a page of confident criteria written six months ago is now mostly wrong.

Finding out for your own set

Run your actual buying questions and record, per question and per date, whether a summary appeared and which sources it cited. Two columns. That is a more reliable guide to your category than any general list, because it is your category.

Do not check once and file it. The set of triggering questions is one of the faster-moving things on this surface, and a snapshot ages badly.

Why it matters commercially

Because it tells you where a click is still available. A question that reliably triggers a summary is one where citation is the goal and ranking alone is not enough. A question that does not is one where the ordinary result still does the work — and those are worth knowing separately rather than averaged together.

questions

The short answers

When does Google show an AI Overview?

For questions a paragraph can usefully finish: definitional, how-to, many comparative and most local queries. It rarely appears for navigational or branded searches. The threshold moves without announcement, so treat any published list as a prior rather than a rule.

Do branded searches get an AI answer?

Rarely. Somebody typing your name already wants you, and a summary adds little, which is why branded traffic is usually the least affected part of a site.

How do I find out which of my queries trigger one?

Run your actual buying questions and record two things per question per date: whether a summary appeared and which sources it cited. Your own set is a better guide to your category than any general list.

Does the answer change by location?

Substantially, for anything with a geography. A local question resolves against where the request appears to come from, which is also why checking from your own machine tells you about your own street.

Why does the same query sometimes show one and sometimes not?

Because the threshold is not fixed and the underlying pages change. Inconsistency is the honest description, which is why this is worth re-checking rather than filing.

Find out in under ten minutes

One free check, no account and no card. You get the answers the assistants gave, not a score.

Every figure quoted here comes from a real question put to a real assistant, saved with the date and the assistant that answered it. This piece was written to answer the search "when does google show ai overview".

free scan

are you in the answer?

Five real questions against a live answer engine with web search on, each asked ten times. That is fifty real calls, so the report lands in under ten minutes rather than while you wait. You see every question before anything runs, and you get the answer text, who got named and the sources the engine read, not a score.

your site/what you sell/where your buyers are/the questions