Will structured data improve our rankings?

Not directly. Google has consistently said it is not a ranking factor, and our own experience matches that. What it can do is make a page eligible for certain result types, describe your entity consistently, and give machine readers unambiguous facts. Treat it as clarity, not as leverage.

Is FAQ markup still worth adding?

For rich results on a commercial site, no — since 2023 those only appear for authoritative government and health domains. We keep FAQPage markup anyway because it structures questions and answers in a form assistants and AI search can extract cleanly, which is a different and increasingly relevant benefit. Just do not add it expecting to see anything in the SERP.

JSON-LD or microdata?

JSON-LD, in a script tag. It is Google's stated preference, it keeps the description in one reviewable block rather than scattered through the markup, and it survives template changes that would otherwise silently break attribute-based microdata.

Does it help with AI search and assistants?

It helps them read you correctly, which is not the same as guaranteeing they cite you. A system extracting facts benefits from explicit entities, dates and prices rather than inferring them from layout. Nothing about schema forces an assistant to use your page — but ambiguous pages are reliably the ones that get summarised wrongly.

Structured Data, and What It Still Earns

Schema markup describes your page in a form a machine can read without guessing. Most advice about it is several years stale — the rich results people are told to chase have largely been withdrawn, while the reason to keep the markup has quietly shifted to a different audience entirely.

Last updated: August 14, 2026

Start your project Get a free estimate

What it is, and what it is not

Structured data is a block of JSON-LD in the page that states plainly what the page is about: this is an organisation, this is its address, this article was updated on this date, this product costs this much. It is a description, not an instruction — search engines are free to ignore it, and they frequently do.

It does not make you rank higher. Google has never treated it as a ranking factor and still does not
It cannot make a page eligible for something the content does not support — marking up a review that does not exist is a manual-action risk, not a shortcut
JSON-LD in a script tag is the format to use; microdata scattered through markup is legacy and harder to keep correct
It is machine-readable meaning: the same facts your page states in prose, restated where a parser cannot misread them
Getting it wrong is usually worse than omitting it, because invalid markup is simply discarded and the effort is wasted

Most of the rich results people chase are gone

This is the part worth updating your assumptions on. Google withdrew or heavily restricted several of the result types that older guidance still recommends chasing, and FAQ markup is the clearest example — since 2023 it has only produced rich results for authoritative government and health sites. Everyone else keeps valid markup that renders nothing.

FAQ rich results: restricted to government and health domains, so on a commercial site the markup earns no visible enhancement
HowTo rich results: withdrawn from desktop and mobile results
We can measure this on our own domain — across three months, Search Console reports exactly one enriched search appearance, on a site carrying FAQPage markup throughout
Still earning their place: Organization and LocalBusiness, Breadcrumb, Product with real price and availability, Article dates, JobPosting and Event
The practical rule: add markup because it states something true and useful, not because you expect a snippet to appear

The audience that changed

The reason to keep structured data well-formed is no longer mainly Google's snippets. It is that a growing share of the systems reading your site are not rendering it for a human at all — they are extracting facts to answer a question somewhere else, and unambiguous markup is the difference between being quoted correctly and being paraphrased wrongly or skipped.

Assistants and AI search parse pages to extract entities, dates, prices and relationships — the exact things schema states explicitly
A visible 'last updated' date carries more weight when a machine-readable dateModified agrees with it
Organization markup with consistent identifiers helps a system decide that the company on your site is the same one it read about elsewhere
Product data that is unambiguous is more likely to be summarised accurately than data inferred from page layout
None of this is guaranteed to be used — but ambiguous pages are reliably the ones that get it wrong

The mistakes we see most

Almost every broken implementation we look at fails in one of a few ways, and all of them are visible in a validator in under a minute.

Markup describing something the page does not show — the fastest way to lose eligibility entirely
Duplicate conflicting nodes: two Organization blocks with different names, so a parser cannot tell which is authoritative
@id references pointing at nothing, leaving a graph of disconnected fragments instead of one description
Dates that contradict the visible page, which undermines the freshness signal it was added to provide
Markup injected client-side by a tag manager, which is fragile and often simply not seen

FAQ

Frequently Asked Questions

Will structured data improve our rankings?
Not directly. Google has consistently said it is not a ranking factor, and our own experience matches that. What it can do is make a page eligible for certain result types, describe your entity consistently, and give machine readers unambiguous facts. Treat it as clarity, not as leverage.
Is FAQ markup still worth adding?
For rich results on a commercial site, no — since 2023 those only appear for authoritative government and health domains. We keep FAQPage markup anyway because it structures questions and answers in a form assistants and AI search can extract cleanly, which is a different and increasingly relevant benefit. Just do not add it expecting to see anything in the SERP.
JSON-LD or microdata?
JSON-LD, in a script tag. It is Google's stated preference, it keeps the description in one reviewable block rather than scattered through the markup, and it survives template changes that would otherwise silently break attribute-based microdata.
Does it help with AI search and assistants?
It helps them read you correctly, which is not the same as guaranteeing they cite you. A system extracting facts benefits from explicit entities, dates and prices rather than inferring them from layout. Nothing about schema forces an assistant to use your page — but ambiguous pages are reliably the ones that get summarised wrongly.