Content negotiation
Cut what an AI pays to read your page.
Read →Agent view — served from /schema-drift.md.
Guide · updated September 2026
What is schema drift, and why does it cost you citations?
Schema drift is when your structured data no longer matches what the page displays — a JSON-LD price of $49 on a page showing $79, or an availability flag that says in stock when it is not. It matters because contradictions undermine an engine’s confidence in everything else you publish, and almost no audit checks for it.
Why does accuracy matter more than volume?
A 2026 Ahrefs study of 1,885 pages found that adding structured data alone barely moved citation rates. That result surprised people who had been told schema was the lever. The likely explanation is that engines validate structured claims against the rendered page, so schema is a trust signal rather than a ranking input — and a signal only helps while it is true.
Which reframes the work. Adding more schema types to a page whose existing markup already contradicts itself makes things worse, not better. Correctness first, coverage second.
What causes schema drift?
| Cause | How it happens | Frequency |
|---|---|---|
| Hardcoded markup | Schema written once by hand, page updated by a CMS | Very common |
| Separate data sources | Schema built from a feed, page rendered from a database | Common |
| Caching mismatch | Structured data cached at a different TTL from the page | Common |
| Template inheritance | A parent template injects defaults a child page overrides visually | Common |
| Migration residue | Old schema left behind after a redesign | Occasional |
Ecommerce suffers most, because prices and stock levels change constantly and the markup is rarely wired to the same source. But the pattern that catches B2B sites is simpler: an Organization block written three years ago listing an address, phone number or founder that has since changed.
Which fields drift most?
- price and priceCurrency — the most consequential, because an agent comparing options quotes them directly.
- availability — recommending an out-of-stock item damages the engine’s credibility as well as yours.
- datePublished and dateModified — frequently automated to today’s date, which is a false freshness claim.
- aggregateRating — often frozen at a number that no longer reflects the reviews shown.
- address and telephone — the quiet B2B one. Wrong for years, noticed by nobody.
Automated dateModified deserves particular attention. Stamping every page with today’s date to look fresh is a claim that is trivially checkable against your actual content, and getting caught making it is worse than being honestly stale.
How do I detect schema drift?
Validators will not find it. Google’s Rich Results Test and schema.org’s validator check whether your markup is well formed, not whether it is true — a perfectly valid price that is completely wrong passes both.
Detection requires comparing the structured value against the rendered value on the same page, field by field. That is straightforward to automate for exact matches such as price and availability, and needs semantic comparison for descriptive fields, which is why we score the exact half deterministically and report the descriptive half as an advisory flag.
$ curl -s https://yoursite.com/product | grep -o '"price":[^,]*' # then open the page and compare
Where this is contested
We cannot prove drift reduces citations. The Ahrefs finding shows schema volume does not increase them, which is suggestive but not the same claim. Our reasoning is that engines validate what they can and weight sources they can verify — plausible, and not established.
What is not arguable is that publishing a wrong price is bad regardless of any AI consideration, and that most sites have no process that would ever catch it. That alone justifies the check.
Questions
Frequently asked
What is schema drift?
When structured data stops matching the visible page: a JSON-LD price that differs from the displayed price, an availability flag that contradicts the page, or an address that changed years ago. Validators do not catch it because the markup is still well formed.
Does more schema markup increase AI citations?
Not on its own. A 2026 Ahrefs study of 1,885 pages found schema volume barely moved citation rates. Accuracy appears to matter more than coverage, because engines validate structured claims against the rendered page.
Will Google Rich Results Test find schema drift?
No. It validates whether markup is well formed, not whether it is true. A syntactically perfect price that is completely wrong passes every standard validator.
Which schema fields drift most often?
Price, availability, dateModified, aggregateRating, and organisation address and telephone. Ecommerce sites drift most on price and stock; B2B sites most often carry outdated organisation details nobody has looked at in years.
See how AI reads your site
Eight checks against your domain in about fifteen seconds. No email.