Google stopped showing FAQ rich results in May 2026 and later removed the documentation. Plenty of sites, mine included, kept the markup anyway on the reasoning that AI systems read structured data. That reasoning is untested, and it is the kind of thing an industry repeats until it becomes received wisdom. So here is the test.
The question
Now that FAQ rich results are gone from Google’s search results, does FAQPage markup still produce any measurable benefit? Specifically, does a page with a visible FAQ plus FAQPage markup get cited by AI systems more often than a matched page with the same visible FAQ and no markup at all.
Why this matters
If the markup does nothing, it is harmless clutter that still has to be maintained, validated and kept in step with the visible content. If it does something for AI citation, it stays, and it stays for a stated reason rather than out of habit. Either answer is useful and the current position is neither.
The method
Sixteen pages across sites I control, all with a visible FAQ section. Eight keep their FAQPage markup, eight have the markup removed while the visible FAQ stays exactly as it is. The visible content does not change on any page, which is the whole design.
| Element | Setting |
|---|---|
| Sample | 16 pages, 8 with markup and 8 with it removed |
| Change | JSON-LD only. The visible FAQ is untouched on every page |
| Baseline | Citation and appearance capture before removal |
| Measurement | The FAQ questions themselves, run weekly on Google AI Overviews, ChatGPT and Perplexity |
| Repeats | Three runs per query per capture |
| Also tracked | Search Console impressions and clicks for each page |
| Duration | 10 weeks |
Removing markup rather than adding it is deliberate. The pages already have history, so there is no new-page indexing delay confusing the result.
What I expect to happen
My prediction, written before the result: no measurable difference in AI citation between the two groups, on any of the three platforms. My honest view is that keeping FAQPage markup after the rich result disappeared is a reasonable-sounding habit rather than a working signal, and I expect this test to show exactly that.
I am fairly confident, which is a good reason to write the prediction down where it can be checked against what actually happens.
Results
The result is pending until the ten weeks are up. When it fills, it will carry the before and after citation counts for both groups across all three platforms, the Search Console impressions and clicks either side of the change, and the run-to-run variance so you can judge the noise.
[PAUL TO ADD: the figures, once the test has actually run.]
What I concluded
The conclusion is pending and follows the data. If removing the markup makes no difference, my honest recommendation becomes that FAQPage is optional rather than advisable, and the service pages saying otherwise get corrected. If it does make a difference, that is a genuinely useful finding.
How to repeat this
Running this needs pages that already have visible FAQs and FAQPage markup, plus a willingness to remove markup from live pages for ten weeks. Setup is an afternoon. The step that matters is leaving the visible content completely alone, because changing it tests something else entirely.
- Find at least sixteen pages with a visible FAQ and valid FAQPage markup. Split into two matched groups.
- Capture a baseline on the FAQ questions themselves across your chosen platforms, three runs each.
- Remove only the JSON-LD from group B. Change nothing a visitor can see.
- Validate that group A’s markup is still valid and still matches the visible content.
- Recapture weekly for ten weeks, and pull Search Console for both groups either side.
Limitations
Sixteen pages on my own sites is small, and AI citation depends heavily on factors this does not control. Removing markup also cannot be undone cleanly in terms of history, so a null result is more trustworthy here than a positive one. Google may also reinstate or change FAQ handling at any point.
FAQ
Did Google really stop showing FAQ rich results?
Yes. FAQ rich results stopped appearing in Google’s search results from May 2026, and the supporting documentation was taken down some time afterwards. The old documentation URL now redirects to Google’s general search updates page, which is where the change is formally recorded.
Should I remove FAQ schema from my site?
Not yet, and not because of this test until it has run. Valid markup that matches your visible content is not doing harm. The question is whether it is doing good, which is what this checks. Removing markup is an easy decision to reverse.
Does structured data help with AI search generally?
Accurate structured data helps search engines understand what a page is about, which supports the clear understanding AI answers depend on. Google has said no special markup is needed for AI Overviews. Those two statements sit together awkwardly, which is why testing individual types is worthwhile.
What about other schema types that were retired?
Several structured data features were retired in June 2025, with more phased out that November. The same question applies to each of them, and the honest answer for most is that markup describing your page accurately is worth keeping even when no rich result follows.
Sources
- Google Search Central, search updates page, recording that FAQ rich results stopped showing from May 2026 and that the FAQ documentation was subsequently removed. Accessed 16 September 2026.
- Google Search Central, AI features documentation, on there being no additional markup requirements for AI Overviews or AI Mode. Accessed 16 September 2026.
Not sure what your schema is doing?
Most sites carry markup nobody has looked at since the day it was added, and a surprising amount of it describes something the visitor never actually sees. That mismatch between the markup and the page matters considerably more than whether any individual type is present or absent.
Get in touch, or start with a free assessment. More tests are on the tests hub.