What Is Schema Markup in SEO?

Schema markup is structured data — a small block of code, almost always JSON-LD, that labels what the content on a page means so search engines don't have to guess. It is not a ranking factor. What it does is make a page eligible for rich results in Google and easier for AI answer engines to parse, extract and quote.

What schema markup actually is

A human reading "4.6 out of 5, 212 reviews, $19/month" knows instantly what those numbers are. A crawler sees three unlabelled strings.

Schema markup fixes that by attaching machine-readable labels. It uses the shared vocabulary at schema.org — a collaboration between Google, Microsoft, Yahoo and Yandex — which defines types like Article, Product, FAQPage, Recipe and LocalBusiness, each with its own properties.

You write it as a JSON-LD block in the page's HTML. Nothing renders. It sits there purely as an annotation: a BlogPosting type declaring the headline, the publication date and the author, for instance. That is the whole idea. Everything else is choosing which types to use and filling them in accurately.

It is not a ranking factor — and that is fine

Google has been consistent on this for years: structured data does not directly improve your position. Adding Article schema to a page will not move it from position nine to position four.

What it changes is eligibility. Rich results — star ratings, FAQ dropdowns, recipe cards, event listings, breadcrumb trails, product prices — are only rendered for pages that declare the corresponding schema. No markup, no rich result, regardless of how well the page ranks.

The knock-on effect is real but indirect: a result occupying more vertical space, with a rating and a price attached, tends to earn a higher click-through rate than a plain blue link at the same position. You are not ranking higher; you are taking up more of the screen. Treat schema as a click-through-rate and eligibility lever rather than a ranking lever and you will set expectations correctly.

The schema types worth your time

Most sites need three or four types, not thirty.

Type	Use it on	What it can unlock
`Organization`	Site-wide, usually the homepage	Entity recognition, knowledge panel data, logo
`BlogPosting` / `Article`	Every guide and blog post	Author, dates, top-stories eligibility
`BreadcrumbList`	Any page inside a hierarchy	A breadcrumb trail instead of a raw URL
`FAQPage`	Pages with genuine Q&A sections	Expandable questions; clean pairs for AI engines
`Product` + `Offer`	Product and pricing pages	Price, availability and rating in the result
`LocalBusiness`	Location pages	Hours, address, map integration
`HowTo`	Step-by-step guides	Numbered step extraction

Two notes on FAQ schema. First, Google narrowed FAQ rich results in 2023 to mostly government and health sites, so the visible dropdown is no longer the reason to add it — the reason is that it hands every answer engine a clean, pre-separated question-and-answer pair. Second, it only belongs on pages where the Q&A is genuinely visible to readers. Marking up questions that don't appear on the page is a violation, not a shortcut.

Where schema and AI answers overlap

This is the part that has changed most recently, and it is worth being precise rather than hopeful about it.

ChatGPT, Perplexity, Claude and Gemini do not need schema to read your page — they parse the rendered text perfectly well. Schema is not a magic ticket into an AI answer, and anyone selling it as one is overstating things.

What it does is remove ambiguity at exactly the points where ambiguity is expensive. A price marked up as an Offer with a currency is unmistakable. A date marked up as datePublished settles the freshness question these systems weight heavily. A FAQPage block presents your answers already split into the shape a model wants to quote.

The bigger lever for AI citation is still the page itself — a direct answer near the top, self-contained paragraphs, real data — which is covered in how to do SEO for ChatGPT. Schema is the cheap, structural half of the same job. So is a well-formed llms.txt file, with the same caveat: useful, low-cost, and not a substitute for the content.

How to add it without a plugin

If you control your templates, do this directly. It takes an afternoon.

Pick the type per template, not per page. All your blog posts get BlogPosting; all your product pages get Product. Generate the values from data you already have.
Write JSON-LD, not microdata. Google explicitly prefers JSON-LD, and it keeps the markup out of your HTML body where a design change can't break it.
Put it in a script tag with type="application/ld+json". Head or body both work.
Use @graph when a page needs several types at once — an article plus its breadcrumbs plus the publishing organisation — rather than stacking separate script tags.
Validate before shipping, with Google's Rich Results Test and the Schema Markup Validator. The first tells you what Google will actually use; the second tells you whether the syntax is valid at all.
Re-check after any template change. Structured data breaks silently — nothing on the page looks wrong when it does.

On WordPress, Yoast and Rank Math generate a reasonable baseline automatically. Validate what they output rather than assuming it is right; the defaults are often generic.

The mistakes that get rich results revoked

Google issues manual actions specifically for structured data abuse, and they are easy to avoid.

Marking up content that isn't on the page. The single most common violation. If the FAQ isn't visible to a reader, it can't be in FAQPage.

Self-serving review markup. Reviews you wrote about yourself, or aggregate ratings that don't correspond to real, visible reviews on the page. This one gets caught often.

The wrong type for the content. A blog post marked up as a Product because product results look better.

Stale values. Prices, availability and dates that no longer match what the page says. A contradiction between your markup and your visible content is worse than having no markup at all.

Marking up a whole page as one thing. A category listing is an ItemList, not a single Article.

The rule underneath all five: schema describes what is already there. It never adds a claim.

Frequently asked questions

Does schema markup improve rankings? No. Google has stated repeatedly that structured data is not a ranking factor. It makes pages eligible for rich results and helps machines interpret your content correctly, which can lift click-through rate — but the position itself is decided by other signals.

What is the difference between schema markup and structured data? Structured data is the general concept of machine-readable labelled data. Schema.org is the specific shared vocabulary nearly everyone uses, and "schema markup" means implementing that vocabulary on your pages. In practice the terms are used interchangeably.

Should I use JSON-LD, microdata or RDFa? JSON-LD. Google recommends it, it lives in a single self-contained script block, and it survives the template and design changes that would break inline microdata attributes.

Does FAQ schema still do anything? Yes, though not what it used to. Google restricted the visible FAQ rich result to a narrow set of sites in 2023, so expect no dropdown. It remains worth adding because it gives answer engines cleanly delimited question-and-answer pairs to quote.

How long until schema markup shows an effect? Google has to recrawl the page first, which typically takes days to a few weeks. Rich results then appear at Google's discretion — valid markup makes you eligible, it does not guarantee the display.

Can bad schema markup hurt my site? Yes, in two ways. Invalid markup is simply ignored, which costs you nothing but the effort. Deceptive markup — invisible FAQs, fake ratings, mismatched types — can earn a manual action that removes your rich results entirely.

The bottom line

Schema markup is plumbing, not strategy. It will not rescue a page nobody wants to read, and it will not lift you up the results page on its own. What it does is make a good page legible: eligible for richer search results, unambiguous to AI answer engines, and correctly attributed to you as a publisher. Add Organization, Article, BreadcrumbList and — where the Q&A genuinely exists — FAQPage, validate them, then move on to the work that actually decides rankings: matching intent, earning links, and being worth citing. The trust side of that equation is covered in E-E-A-T in SEO.

Structured data makes a page readable. Links are what make it rank. Backlinkster handles the second half — one-for-one in-content link swaps with owners of related sites, each verified live by code, five a month on the free plan. See the plans.