Google removed FAQ rich results for most sites in May 2026. Seven structured data types were retired outright in June 2025, including Course Info, Claim Review, Estimated Salary and Vehicle Listing, and a further round was announced for January 2026. John Mueller has been clear that Google is not killing schema and that rankings are unaffected.
Both halves of that matter. A visual feature was withdrawn, so the expanded list of questions under your result is not coming back. The markup underneath it still does work, and stripping it out on the assumption that it is now pointless is the wrong conclusion.
What actually changed
The distinction people miss is between a schema type being invalid and a rich result being withdrawn.
When Google retires a feature, the schema.org type usually remains perfectly valid. Your markup still parses, still validates, and still causes no harm. What stops is the visual treatment in the results page.
Nothing about your ranking moves. If you saw a traffic drop around one of these dates, it came from losing the extra vertical space your result occupied and the clicks that space attracted, not from a demotion. Those are different problems with different responses, and confusing them sends teams looking for a penalty that does not exist.
The practical test is whether impressions held while clicks fell. That pattern is a lost visual feature. Impressions falling is something else entirely.
Why the markup is still worth having
Three reasons survive the removal of the snippet.
Assistants and language models retrieve from structured content. A question and answer pair marked as such is easier to extract cleanly than the same text as ordinary prose, and extraction is how you get cited rather than summarised past.
Search still uses it to understand the page even where it does not decorate the result. Telling a machine that this block is a question and this block is its answer is useful regardless of how the result renders.
And the discipline it imposes is worth more than the markup. Writing content that can be marked up as questions and answers forces self-contained answers, which is the property that makes a page useful to a reader arriving mid-page from anywhere.
We kept FAQ markup across our own posts after this change for exactly those reasons, and adjusted the expectation rather than the implementation.
Where rich result effort belongs now
If you were investing in schema for the visual return, redirect it rather than abandoning it.
Product and Review markup still earn rich results and still affect how a listing looks against competitors. For anyone selling something, that is where the remaining visual upside sits.
Core types describing what a page is, who published it and when remain foundational. Article, Organisation, Breadcrumb and the basic identity markup are not decorative features and are not going anywhere.
What is not worth chasing is the next rich result. The pattern over several years is that Google adds visual treatments, watches how they are used, and withdraws the ones that are gamed or that reduce clicks to sites. Building a content strategy around a specific snippet type is building on something designed to be revisable.
Audit what you are still emitting
A short, unglamorous task worth doing once.
List the schema types your site outputs and check each against the retired list. Markup for a withdrawn type is harmless and it is also dead weight in your templates that somebody will eventually maintain without knowing why.
Check for errors while you are there, because invalid markup is a genuinely different situation from retired markup. The first can suppress results you would otherwise get; the second simply does nothing.
Then stop reporting on it as a channel. If a dashboard has a panel counting FAQ rich results, that panel is now measuring zero forever, which is the same flat-line failure we described in a metric you report on has been deleted.
When we run this audit the common finding is not retired markup at all. It is two different schema blocks describing the same page, added years apart by different plugins, disagreeing with each other about the publication date.
The pattern is worth learning from
This is the second time in a year that a measurement or display feature has been withdrawn under sites that were reporting on it.
The transferable lesson is to be clear about which of your investments depend on a third party continuing to render something. Content that is genuinely useful survives any change to how results are displayed. Content built specifically to trigger a visual feature has a dependency with no contract behind it.
That is the same concentration risk we described in where demand comes from now, expressed as a feature rather than a channel.
Our website development work treats structured data as something that helps machines understand a page rather than as a route to a particular snippet, which is why this change required an adjustment of expectation and no code.
What to do this week
Three things, none of which takes long.
Check whether your traffic pattern shows steady impressions with falling clicks around the change, so you know whether this affected you at all. Remove or ignore any dashboard panel now permanently at zero. Audit your emitted types against the retired list and clear out anything dead.
Then leave the FAQ markup in place. It costs nothing, it helps machines read the page, and the writing discipline behind it is the part that was always doing the work.
Frequently Asked Questions
Should we remove our FAQ schema?
No. The type is still valid, it helps machines parse the page, and it aids extraction by assistants. Only the visual rich result was withdrawn. Adjust the expectation, not the markup.
Did this affect rankings?
No. Rankings are unaffected. Any traffic change comes from losing the extra vertical space your result occupied and the clicks that attracted, which is a display change rather than a demotion.
How do we tell if we were affected?
Compare impressions against clicks around the change. Steady impressions with falling clicks indicates a lost visual feature. Falling impressions is a different problem entirely.
Which schema types still earn rich results?
Product and Review remain the ones with visual return for most commercial sites. Core identity types such as Article, Organisation and Breadcrumb remain foundational regardless of display.
Is retired markup harmful?
No, just inert. The schema.org type usually stays valid and your markup still parses. It is dead weight in a template rather than a risk, which is different from invalid markup.
Should we chase the next rich result type?
Cautiously. Google adds visual treatments and withdraws the ones that get gamed or reduce clicks. A content strategy built around one snippet type has a dependency with nothing behind it.
What about structured data for AI search?
It remains useful, and it is now the stronger argument for keeping it. Clean question and answer pairs are easier to extract accurately than the same content as prose.
How often should we audit our schema?
Once a year, plus whenever a change is announced. Check emitted types against the retired list and check for validation errors, which are a genuinely different problem from retirement.
Talk to our team if your clicks fell while your impressions held and nobody can explain it.

