You put the page into the Rich Results Test, everything is green, but in Search Console the "Items with warnings" report — or worse, "Invalid items" — is climbing, and the rich results you used to have are fading away one by one. This almost always follows a recurring pattern: a JSON-LD block has been placed in the site's shared template (header or footer) that repeats on every page, and that same block, on a page where it shouldn't be, claims a type of structured data that doesn't match that page's content.
First of all, separate out which type of structured data belongs to which page. This decision — not the coding — is the hard part.
Which schema for which page type
A simple rule that works in practice: choose the schema based on the page's visible content, not on what you hope Google will display. If the data isn't visible on the page, you have no right to put it in JSON-LD.
| Page type | Suitable schema | Unsuitable schema |
|---|---|---|
| Blog article | Article or BlogPosting | Product, FAQPage without real questions |
| Product page | Product + Offer | Article |
| FAQ page | FAQPage | Article |
| All internal pages | BreadcrumbList | — |
| Homepage | Organization or WebSite | Product |
A point that's less often seen: Article and Product can coexist on one page, but only when both genuinely exist. A page that is both a product review article and has a buy button can provide both; a page that is just text doesn't get Product.
Article: the fields that actually matter
The minimum set Google wants for Article is headline, image, datePublished, and author. But what actually causes warnings is the dates. If you've hardcoded a fixed dateModified in the template, it must change after every content edit; otherwise Google sees a false modification date.
{
"@context": "https://schema.org",
"@type": "Article",
"headline": "Structured Data: Which Schema for Which Page",
"image": ["https://example.com/img/cover.jpg"],
"datePublished": "2025-03-11T09:00:00+03:30",
"dateModified": "2025-03-11T09:00:00+03:30",
"author": {"@type": "Person", "name": "Author Name", "url": "https://example.com/author/x"}
}
Provide dates with the Tehran timezone. Don't drop +03:30; without it, Google assumes the time is UTC and may read the publication date as a day earlier.
Product: where your price gets you into trouble
Product without offers is almost useless. If you put the price in the schema, it must match the price visible on the page. On Iranian sites where the price changes based on the daily rate or includes tax, this synchronization is a constant source of error. If you can't inject the price from the database into JSON-LD in real time, it's better to provide offers with priceCurrency: "IRR" and a base price and show the same on the page, or skip the price rich result entirely. In this case I choose the second option: a wrong price rich result is worse than not having one.
FAQPage: a limitation many don't know about
Since August 2023, Google has limited the display of FAQ rich results to government and health sites only. That means if you put FAQPage on an online store, the data is valid but the chance of seeing a visual result in the SERP is low. That's not a reason to remove it; it's still useful for machine understanding of content, but don't expect extra clicks.
The more common problem is this: you put FAQPage in the shared template and the same three generic questions ("business hours," "shipping method," "warranty") repeat across a thousand pages. Google sees this as duplicate content at the structural level. FAQPage only makes sense on a page whose questions are specific to that page.
BreadcrumbList: the cheapest and least troublesome
If you only have time for one schema, choose BreadcrumbList. It carries the least risk and shows the path in the SERP. Just make sure the breadcrumb path in the schema matches the actual URL path; if you've written /blog/seo/ in the schema but the actual URL is /mag/seo/, you'll get a warning.
The rule whose violation knocks the whole domain out of rich results
I say this with confidence because I've seen it many times: if a schema type is invalid across a significant portion of a site's pages, Google can reduce its trust in that type across the entire domain. A broken Product block in the shared template repeats across a thousand pages and shows up in Search Console as a sudden spike of "Invalid items." That's when healthy pages also lose their rich results, and you go looking for the cause in the wrong place.
The structural solution, not a patch: don't put JSON-LD in the shared template. Inject it at the level of each template separately, and for each page type, provide only the type that belongs to it.
This is where they go wrong
A common mistake I see in site audits: placing several separate JSON-LD blocks on one page, each with its own @context, defining a duplicate Organization. On the surface you get no error, but Google sees two different entities with the same name and may not recognize either as the site's primary entity. The sign is this: in the Knowledge Panel or in brand results, the logo or contact information isn't displayed correctly, while in the Rich Results Test everything is green.
The right way: build a single block with @graph and connect the entities with @id.
{
"@context": "https://schema.org",
"@graph": [
{"@type": "Organization", "@id": "https://example.com/#org", "name": "..."},
{"@type": "WebSite", "@id": "https://example.com/#site", "publisher": {"@id": "https://example.com/#org"}},
{"@type": "Article", "publisher": {"@id": "https://example.com/#org"}}
]
}
The second mistake I see a lot: copying the sample code from Google's documentation without changing the url and @id. The result is that several pages of your site reference the same address, and Google recognizes the homepage as the author or publisher of all articles. For a quick check of these kinds of errors, the SEO and webmaster tools available to you are enough; compare the output with the page's actual URL.
How to validate without breaking the site
- First, open the Enhancements section in Search Console and see which schema type has the most errors. Start with the most frequent, not the newest.
- Put a sample page from each template into the Rich Results Test. If it has errors, fix that one page manually before making any change to the template.
- After fixing, count how many JSON-LD blocks are on the page with
curl -s https://example.com/page | grep -o 'application/ld+json' | wc -l. If there's more than one and each defines Organization, fix that right there. - Apply the change on a subdomain or staging and wait 48 hours for the Search Console data to update.
A technical point that eats up a lot of teams' time: if the site is behind a CDN or page cache, old JSON-LD may be served hours after new content is published. After any change to a date or price, purge the page cache; otherwise the structured data will be inconsistent with the actual content. To make sure the version Google sees is the same version you see, check the domain and DNS technical check and the response headers.
If the SEO team and the development team are separate, explicitly assign responsibility for JSON-LD to one of them. My experience is that when no one owns it, both sides assume the other is maintaining it, and six months later no one knows where that block in the template came from. If you're outsourcing this, ServerNet's SEO services is one option, but in any case, ask that the technical ownership of the code stay in your own repository.
For implementation patterns and ready-made examples, the documentation and knowledge base is a good starting point. And if you want to know how structured data interacts with real page speed, read Measuring real user performance with CrUX; a rich result on a page that doesn't load in three seconds has little practical value.
Frequently Asked Questions
Does structured data directly affect rankings?
The schema itself is not a direct ranking signal. What it does is help Google understand the page content more precisely and increase the chance of appearing in rich results. That display can change the click-through rate, and that's the indirect path.
Why did my site's rich result suddenly disappear from the SERP?
The most common cause is a schema type becoming invalid in the shared template that has repeated across a large number of pages. First check the Enhancements section in Search Console and see which type has the most errors; usually that one is responsible.
Is it allowed to have multiple schema types on one page?
Yes, provided each type matches the actual, visible content of that page. Article and BreadcrumbList almost always make sense together. But only add Product when the page is genuinely a product page with a visible price.
Is JSON-LD better or Microdata?
Choose JSON-LD. It's separate from the HTML, easier to maintain, and Google also presents it as the recommended format. Microdata only makes sense if for some reason you can't add a script, which almost never happens.
Next step: right now, put one page from each of your site's templates into the Rich Results Test and see how many JSON-LD blocks it has. If there's more than one and each defines Organization, fix that today.
Comments 0
No comments yet — be the first!