Featured snippet readiness: what has to be true all at once
There is no featured-snippet tag, no schema type and no setting. Position zero is an outcome, awarded when several independent things happen to be true of one page at the same time. This check reports whether they are, which is why it is the only composite in the catalogue.
The short answer
Featured snippet eligibility is a set of conditions that have to hold at once: the page ranks in the top ten, it contains a passage that directly answers the query, that passage is in the right format for the question, it is marked up properly, and it sits near a matching heading. It is a readiness checklist, not a trick.
What has to be true at the same time
Five conditions, and the important property is that they are conjunctive. Failing any one of them takes the page out of contention regardless of how well it does on the other four, which is why isolated snippet advice so often produces no result.
- You already rank on the first page. Snippets are drawn from results that are already competitive. This is the condition most snippet guidance omits, and it is the one that makes the rest conditional: if you are on page three, none of the other four items matter yet.
- The query is a question. Informational intent, phrased as something answerable. Navigational and transactional queries do not produce snippets, so half the keywords people target for this were never eligible.
- The page contains a passage that answers it. Standalone, quotable, no unresolved references. This is the direct answer block, and its absence is the most common single reason a well-ranking page never gets lifted.
- The passage is in the right shape. Paragraph, list or table, matching what the question wants. A prose answer to a steps question is competing for a format it cannot fill.
- The question is findable in your structure. A heading that states the question, so the engine can pair the question with the passage beneath it rather than inferring the pairing from the body text.
How this check is calculated
It reads the results of the other four page-level checks for the same URL and reports whether they align, rather than measuring anything itself. That is unusual enough to be worth stating plainly: there is no field on the page that this check inspects.
What it cannot see is the first condition. Whether you rank is not a property of your markup, and no crawler knows your position for a query it was never given. So the honest output is conditional: this page is structurally ready, or it is not, and whether it ranks is yours to check in Search Console.
A structurally perfect page ranked eleventh wins nothing. Fix the ranking first, then this.
Checking a page against the five
Four of the five are things you can read off your own page in a couple of minutes: is there a question-shaped heading, is there a standalone passage under it, is it the right shape, and does it answer without pointing elsewhere.
This one needs the full crawl
This one is a composite: it reads the results of several other checks for the same page and reports whether they align. There is no single field to measure, which is why it only exists inside a full audit.
The instant search on this site audits a single page, so rather than show you a verdict it cannot support, this guide sends you to the place the check actually runs.
One page, before and after
The change that flips a page from ineligible to eligible is usually small and structural rather than a rewrite.
<!-- BEFORE: ranks fourth for "how long should a
crawl take", and has never taken the snippet.
The answer is present but has no question
heading above it and no boundary around it. -->
## Performance
Crawl duration depends on a number of variables that
interact, and it is worth understanding them before
setting expectations with a client. Concurrency is the
main lever, though the origin server's response time
usually dominates in practice. A mid-sized site will
typically finish in well under an hour.
<!-- AFTER: the question becomes the heading, the
answer becomes the passage under it, the detail
stays where it was. Same content. -->
## How long does a full site crawl take?
A crawl of a few hundred pages usually finishes in
under an hour. The origin server's response time sets
the pace more than concurrency does, so a slow host is
the usual reason a crawl runs long.
### What changes the duration
Concurrency is the main lever you control, though the
origin's response time normally dominates…Three edits: the heading became the question, the answer moved above the explanation, and the explanation kept its own subheading. Nothing was deleted, and the page went from having no liftable passage to having one.
Winning one is not permanent, and losing one is not a penalty
Snippets move. The same query can show a different source week to week, and the snippet can disappear entirely when the engine decides the query no longer warrants one. None of that is a signal about your page.
Which has a practical consequence for how you treat this: it is not a metric to hold steady, because you cannot. Fix the structural conditions, then let it be volatile. Chasing a snippet you held last month by making weekly edits to a page that already meets the conditions is effort spent on noise.
The zero-click objection deserves a straight answer: sometimes winning one costs you the visit.
If your page's entire value is a fact someone wanted, and the snippet gives them that fact, they had no reason to click and you have been read without being visited. That is a real trade, not a myth. It argues for answering the question completely and making the page worth opening for the reasoning, the caveats and the specifics that a forty-word passage cannot carry. It does not argue for withholding the answer, which loses you the snippet and the click.
Why a composite is rated Critical
The rating looks odd at first: this is an opportunity check, and its four inputs are each rated Important. A composite of Important checks being Critical needs justifying.
The reasoning is that it is the only check in the answer-engine group that reports an outcome rather than a property, and its failure mode is a page that has done all the hard work and gets none of the return. A page ranking on the first page of results for a question it answers well, losing the box to a page below it, is the most expensive near-miss in the catalogue: the traffic is going to a competitor for a structural reason you can fix in an afternoon.
Severity on this scale is about consequence and reach rather than about how hard something is to fix, and a lost position zero on a page that already ranks is a large consequence.
Does this affect AI search?
The conditions overlap almost completely, which is the most useful thing on this page.
A passage that stands alone, answers early, is shaped to match the question and sits under a heading that states it, is exactly what a retrieval pipeline wants and exactly what a snippet algorithm wants. The two systems are not related and they reward the same structure, because both are solving the problem of finding a quotable answer inside a document.
The parts that do not transfer are worth knowing so this is not oversold. Ranking is not a precondition for being cited by an assistant, which fetches on demand rather than choosing from a ranked list, so a page outside the first page of results can still be quoted. And the snippet formats, paragraph against list against table, are a search-surface concept with no direct equivalent in a generated reply.
So: do this work for the snippet, and take the AI citation as the same work paying twice. That is a rare enough alignment that it is worth prioritising over the checks where the two diverge, and it is the argument the rest of the AI search group builds on.
Confirming eligibility, then watching for the result
Two separate confirmations on two separate timescales. The structural one is immediate: re-crawl and the four inputs should read as satisfied for the page you edited. That is the part you control and the part worth checking off.
The outcome is slower and outside your control. Watch the query in Search Console for a change in position and impressions rather than searching for it yourself, since your own results are personalised and a snippet you can see may not be the one anyone else gets. Give it weeks rather than days, and treat the absence of a snippet on a page that meets all five conditions as normal rather than as a fault to chase.
Questions this check raises
- How do I get a featured snippet?
- Rank in the top ten first, because snippets are drawn from pages that already rank. Then give the query a direct answer in the format the question implies, place it under a heading that matches the question, and mark it up as a real paragraph, list or table. There is no tag that requests one.
- Do featured snippets steal traffic?
- Sometimes, on queries where the answer is all anyone wanted. On queries where the answer is the start of a task, the snippet acts as a qualification: the click that follows is from someone who already knows this page has what they need. Which effect dominates depends on the query, not on the snippet.
- Why did I lose a featured snippet I used to have?
- They are recalculated constantly and are not held. A competitor published a better-shaped answer, the query intent shifted, or Google decided the query no longer warrants one. Losing it is not a penalty and there is nothing to diagnose beyond whether the page still answers the question well.