Google AI Overviews
The block above the results is built from the results.
An AI Overview is assembled from pages Google already returns for the query. That makes the work familiar — rank, answer directly, cite your sources — and makes it unforgiving: the page has to be quotable in a single paragraph.
One query per postAnswers in the first sentenceFigures attributed or dropped
What a summary is actually made of.
Four shapes of answer that get lifted out of a page. Each one is a decision made before writing, not a flourish added after it.
- The definition
- A question whose answer is a short factual statement gets answered above the fold. The page that supplies it defines the thing in its first sentence rather than three paragraphs in, after the throat-clearing.
- The steps
- Procedural questions are summarised as a numbered list. A page whose headings are already the steps, in order, hands the block its structure; a page written as flowing prose gives it nothing to copy.
- The comparison
- Two options, one difference that matters. A table or a plainly stated distinction is the most quotable thing a page can carry, and the hardest thing to write without inventing something a reviewer would have to strike out.
- The threshold
- A number that decides something — a limit, a deadline, a percentage, a price — attributed to the body that set it. This is the part of an answer people act on, and the part that has to be right.
What the page has to do, in order.
Five requirements, four of which are the same ones classic search has always had. The odd one out is the writing: an answer a machine can lift without losing its meaning.
- 01
Rank first
The block is assembled from results Search already returns. A page that ranks nowhere is not a candidate, so the query, the intent and the on-page work still come first — there is no shortcut around being a result.
- 02
Answer in the first sentence
Every section opens with a complete answer of 25–50 words, written to survive being quoted on its own. The explanation follows; it does not come first.
- 03
Carry the specifics
Figures with a named source next to them, dates that are current, and nothing stated that cannot be traced. A block that quotes you once and is wrong will not quote you twice.
- 04
Mark it up
BlogPosting structured data with the date and language, FAQPage where the article answers questions, a canonical URL and a sitemap entry. Markup does not rank anything by itself; it removes the ambiguity about what the page says.
- 05
Keep it current
An answer that quotes last year's threshold is worse than no answer. Posts are dated, and a plan that keeps publishing gives the block newer material to prefer.
Writing for the click, or writing to be the answer.
The two used to be the same job. They are not any more, and the difference is visible in the first sentence of every section.
Writing for the click
- The answer withheld until after the introduction, to buy a scroll
- A headline that promises more than the page delivers — the clickbait tax
- Numbers quoted from other people's blog posts, with no chain back to a source
- One long page covering every related question, so no section answers one cleanly
- The page never revisited, so its figures age in place
Writing to be the answer
- The answer in the first sentence, then the detail that earns the scroll
- A title that states the promise the article keeps
- Every figure attributed, and dropped by the review when it cannot be
- One section per question, so each one can be lifted without the rest
- A dated post on a cadence, because the specifics have a shelf life
You still see what each post earned.
Connect Search Console read-only and the clicks, impressions and average position of the posts you publish sit beside the calendar that produced them. That is where the trade-off becomes a decision you can make with numbers instead of an argument about strategy.
It will not tell you that you were quoted in a summary, because nothing outside Google can. It will tell you whether the page is being found for the query it was aimed at — and if it is not, that is a query problem you can act on.
Being cited by assistants1,204
Clicks
48,910
Impressions
12.4
Avg. position
One thing worth deciding before you start.
Summaries take the click you were counting on, so the question is who you would rather the buyer read.
Fewer clicks, or a competitor's paragraph
A block that answers the question removes the reason to click through. That is a real cost, and it lands hardest on queries whose answer is a fact — a definition, a threshold, a deadline. It lands lightest on commercial queries, where the reader still wants the comparison in their own context and usually arrives at a page that made the shortlist.
The alternative is not the old click-through rate. It is the same answer with somebody else's page inside it. If a buyer is going to read a summary of your category regardless, the only question left is whether your site is one of the sources.
Who still makes the editorial call.
The pipeline writes and checks the page. It cannot decide what your business is willing to state in public — and a summary has no room for the nuance a sales conversation would have carried.
- The numbers you are willing to publish, and the ones you are not
- Which queries you want a summary for, and which you would rather answer yourself
- Claims about pricing, security and compliance that only your team can sign off
- The trade-off you accept: being summarised means fewer clicks, sometimes zero
- Who you allow to be named in a comparison, and how the difference is stated
Related use cases
- GEO for AI answers
GEO without the theatre: one answer per page, quotable and attributable.
- llms.txt
The summary file for language models, and what ours contains.
- For AI search
Be the source the answer cites.
- For multilingual markets
Which market to open next, and what it costs.
Questions about AI Overviews.
Including whether markup helps, and what it costs in clicks.
Does schema markup get me into an AI Overview?
No. Structured data is not a ranking factor on its own, and no markup makes a page eligible for the block. What it does is remove ambiguity: it tells the machine what the page is, when it was published, in which language, and which questions it answers. That helps a system summarise the page correctly — it does not make an unrankable page rankable.
Will being summarised cost me clicks?
Sometimes, yes, and it is worth deciding in advance how you feel about that. If the block answers the question, some readers will not click through. The alternative is being absent from the answer a buyer reads anyway — and for a comparison or a price query, absence usually means the competitor is the one quoted. The plan does not hide this trade-off; the reporting shows what each post actually earned in clicks and impressions, so you can judge it on your own numbers.
How do I find out whether I am being summarised?
Search Console reports the impressions and clicks your pages earn for the queries they appear for. It does not tell you whether a summary was shown, or whether you were part of it. What it does show is whether a page is being found for the query at all — which is the precondition for everything else, and the number Xt4b reports beside your calendar.
Which queries get an AI Overview?
The trigger is decided per query by Google's systems, and it changes: some queries show it consistently, others intermittently. That uncertainty is exactly why the plan does not try to target summaries. It targets the queries your buyers search, and writes the pages to be rankable and quotable — so that where the block appears, the page is already a candidate for it.
Is this just featured snippets with a new name?
The extraction is similar — a direct answer, in the right structure, from a page that ranks — but the output is prose assembled from one or more sources, so the page can contribute without being the single winner the snippet was. The practical consequence for a publisher is the same: answer-first, structured, and sourced is the shape that gets used.
Can you guarantee we end up in one?
No. Nobody can, and the guarantee here is deliberate: Xt4b answers for what it publishes — one query per post, the answer first, sources attached, structured data, a date, a canonical URL on your own domain. Whether Google shows a summary, and whose page it quotes, is Google's decision.
Start with one market
Publish the page the block can summarise.
Connect the domain and pick a language: the first week of entries lands on the calendar, each one aimed at a query and written so its answer can be lifted.
One query per postSearch Console reporting3 days free, no card