Schema, Explained · a three part series for people who know a little code
- Part 1: Organization, WebSite, WebPage, and BreadcrumbList
- Part 2: LocalBusiness, Service, Article, and local search
- Part 3: FAQ, Person, Event, Recipe, and reviews (you are here)
Content-specific schema only works when it describes content that is visibly on the page. FAQPage marks up questions and answers shown on the page; Google now shows FAQ rich results mainly for well-known government and health sites, but the markup still states your answers clearly. Person describes a real individual, ideally once on their bio page with an @id that articles point to. Event needs a name, a start date with a time zone, and a location, plus status and attendance mode. Recipe needs a name and image, and works best with ingredients, steps, times, and yield. Reviews and ratings must reflect reviews shown on the page, and Google ignores self-serving reviews of your own business. When in doubt, ask one question: could a visitor see every fact in this block on the page?
What is the one rule for content-specific schema?
Every type in this post follows the same rule, and it is a content rule, not a code rule: the schema must describe what a visitor can see on the page.
FAQ markup for questions that are not on the page, a five-star rating with no reviews shown, an event date that is different from the one in the hero, a recipe time that does not match the card. Google's guidelines treat these as misleading, and they can cost a page its search features or, in serious cases, lead to a manual action. They also make AI systems less likely to trust anything else you say.
That makes these types the easiest for a creative director to review. You do not need to understand every bracket. You need to hold the page in one hand and the schema in the other and check that they say the same thing.
How do you check FAQPage schema?
FAQPage schema marks up a list of questions and answers that appear on the page.
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "How long does a website audit take?", // the question, word for word as on the page
"acceptedAnswer": {
"@type": "Answer",
"text": "A focused audit for a small-to-mid-size site typically takes one to two weeks." // the visible answer
}
},
{
"@type": "Question",
"name": "What do I get at the end of the audit?",
"acceptedAnswer": { "@type": "Answer", "text": "A prioritized action plan." }
}
]
}Shortened from the FAQ schema on the Double Atari website audits page. The // notes are added for this article. Real JSON-LD does not allow comments, so never paste them into a live page.
What to check:
- Every question and answer is visible on the page, with the same wording. An accordion is fine. Hidden-from-everyone is not.
- The answers are real answers, not "Contact us to find out."
- No advertising. FAQ markup is for genuine questions, not sales copy dressed as questions.
- One FAQPage per page, not the same FAQ block copied onto every template.
What has changed: since 2023, Google has shown FAQ rich results mainly for well-known, authoritative government and health websites. For most businesses, FAQ markup will no longer add expandable questions under a search result. I still recommend it where a page has a real FAQ, because it states your answers in clean question-and-answer pairs, and a well-written FAQ is exactly the kind of passage AI answers draw on. Just do not add an FAQ section only to get the markup.
How do you check Person schema?
Person schema describes a real individual: a founder, an author, a doctor, an attorney, a speaker. It matters more than it used to, because both search engines and AI systems weigh who is behind content.
{
"@context": "https://schema.org",
"@type": "Person",
"@id": "https://www.double-atari.com/about#person", // one name tag, used by every article
"name": "Andrew Charon",
"url": "https://www.double-atari.com/about", // the bio page
"image": "https://www.double-atari.com/images/authors/andrew-charon.jpg",
"jobTitle": "Founder and SEO Consultant", // the same title everywhere
"worksFor": { "@id": "https://www.double-atari.com/#localbusiness" },
"knowsAbout": ["Technical SEO", "Schema markup", "CMS strategy", "Analytics"],
"sameAs": ["https://www.linkedin.com/in/example-profile/"] // the person's own profiles
}Based on the Person block on the Double Atari About page. The LinkedIn URL is a placeholder. The // notes are added for this article. Real JSON-LD does not allow comments, so never paste them into a live page.
| Check | What good looks like |
|---|---|
| Described once | The full Person block lives on the bio page. Articles point to it by @id. |
| jobTitle | The same on every page. Mine currently differs between the homepage and About page, which part 1 flagged. |
| image | A real photo of the person, matching the one on the bio page |
| sameAs | The person's own profiles, such as their personal LinkedIn. Not the company page. |
| knowsAbout | A short, honest list of real areas of expertise, not every keyword you want to rank for |
For a bio or team page whose main subject is the person, you can also use ProfilePage as the page type, with the Person as its mainEntity. Google supports ProfilePage markup for pages about a single person or organization.
How do you check Event schema?
Event schema describes something that happens at a specific time: a workshop, a concert, a webinar, an open house. Google can show events in a special listing in search, so the details need to be exact.
{
"@context": "https://schema.org",
"@type": "Event",
"name": "Website Audit Workshop for Nonprofits",
"description": "A two-hour hands-on session on finding the website fixes that matter most.",
"startDate": "2027-03-04T18:00-06:00", // date, time, and time zone offset
"endDate": "2027-03-04T20:00-06:00",
"eventStatus": "https://schema.org/EventScheduled", // or EventCancelled, EventPostponed, EventRescheduled
"eventAttendanceMode": "https://schema.org/OfflineEventAttendanceMode", // Offline, Online, or Mixed
"location": {
"@type": "Place",
"name": "Example Community Center",
"address": { "@type": "PostalAddress", "streetAddress": "100 Example St",
"addressLocality": "Minneapolis", "addressRegion": "MN",
"postalCode": "55401", "addressCountry": "US" }
},
"image": "https://www.example.com/images/workshop.jpg",
"organizer": { "@id": "https://www.example.com/#organization" },
"offers": {
"@type": "Offer",
"price": "0", // a number, no currency symbol
"priceCurrency": "USD",
"availability": "https://schema.org/InStock",
"url": "https://www.example.com/events/audit-workshop/"
}
}A made-up example. The // notes are added for this article. Real JSON-LD does not allow comments, so never paste them into a live page.
| Check | What good looks like |
|---|---|
| startDate | Matches the page, includes the time, and includes the time zone offset (-06:00 for Central in winter, -05:00 in summer) |
| eventStatus | Updated when plans change. A cancelled event still marked EventScheduled is a common and avoidable mistake. |
| eventAttendanceMode and location | In person needs a Place with an address. Online needs a VirtualLocation with a URL. Mixed needs both. |
| offers | The price and currency match the page. Free events use 0. |
| One block per event | Each event on its own page, or one block per event on a calendar page. Not one block for a whole season. |
One limit worth knowing: Event markup is for events people can attend. A sale or a coupon is not an event.
How do you check Recipe schema?
Recipe is one of the most visually rewarding schema types, because Google can show a photo, ratings, and cooking time right in search. It is also one of the most detailed.
{
"@context": "https://schema.org",
"@type": "Recipe",
"name": "Weeknight Wild Rice Soup", // required
"image": ["https://www.example.com/images/wild-rice-soup-16x9.jpg"], // required, a real photo of the dish
"author": { "@type": "Person", "name": "Example Cook" },
"datePublished": "2026-11-01",
"description": "A Minnesota classic in about an hour.",
"prepTime": "PT20M", // 20 minutes, written as an ISO 8601 duration
"cookTime": "PT40M",
"totalTime": "PT1H", // should equal prep plus cook
"recipeYield": "6 servings",
"recipeCategory": "Soup",
"recipeCuisine": "American",
"recipeIngredient": [
"1 cup wild rice",
"6 cups chicken stock",
"1 cup heavy cream"
],
"recipeInstructions": [
{ "@type": "HowToStep", "text": "Rinse the wild rice and simmer it in the stock for 40 minutes." },
{ "@type": "HowToStep", "text": "Stir in the cream and season to taste." }
]
}A made-up example, shortened. The // notes are added for this article. Real JSON-LD does not allow comments, so never paste them into a live page.
Durations look strange at first, so here is the decoder: PT means "a period of time", then hours and minutes. PT20M is 20 minutes, PT1H is one hour, and PT1H30M is an hour and a half.
| Check | What good looks like |
|---|---|
| image | A real photo of this dish, not a stock photo or a logo |
| Times | Match the recipe card, and total time adds up |
| recipeIngredient | Every ingredient, one per line, with amounts, matching the page |
| recipeInstructions | Each step as its own HowToStep, in order, matching the page |
| Ratings | Only if real visitor ratings are shown on the page |
Recipe sites usually get this from a plugin. The plugin handles the code, but it cannot check the content. A prep time typed as "20" instead of 20 minutes, or a missing photo, is a content problem a reviewer can catch.
How do you check review and rating schema?
Review and rating schema is the most abused type on the web, so Google's rules are strict. There are two kinds:
- Review: one review by one person, with a rating and text.
- AggregateRating: a summary, such as 4.8 stars from 120 ratings.
"aggregateRating": { // added inside a Product, Recipe, Course, etc.
"@type": "AggregateRating",
"ratingValue": "4.8", // matches the stars shown on the page
"reviewCount": "120", // matches the count shown on the page
"bestRating": "5"
}The // notes are added for this article. Real JSON-LD does not allow comments, so never paste them into a live page.
| Rule | What it means for you |
|---|---|
| The reviews must be on the page | If the page shows no reviews, the schema cannot claim any. |
| No self-serving reviews | Google does not show stars for reviews a business publishes about itself in LocalBusiness or Organization markup. Testimonials on your own site will not earn stars. |
| Real reviewers | A Review author should be a real person or organization, not a job title or a marketing label. |
| Right thing being reviewed | Stars belong on the specific product, recipe, course, or book being rated, not on the whole site. |
| Numbers that match | ratingValue and reviewCount match what is shown on the page and are kept current. |
I found one of these on my own site while writing this series, and it is now fixed. I searched the homepage source for "publisher" and found five testimonials where each reviewer's job title and company had been typed in as if they were a publishing organization.
// Before: one of five testimonials
{
"@type": "Review",
"author": { "@type": "Person", "name": "Sarah Abdelaal" },
"publisher": {
"@type": "Organization",
"name": "Engagement Director, Pillsbury House + Theatre" // a job title is not an organization
}
}The // notes are added for this article. Real JSON-LD does not allow comments, so never paste them into a live page.
The fix: a review's publisher is the site or company that published the review, not the reviewer's employer. So I removed publisher and moved the job title and company onto the author, where schema.org has a place for each.
// After
{
"@type": "Review",
"author": {
"@type": "Person",
"name": "Sarah Abdelaal",
"jobTitle": "Engagement Director", // the job title
"worksFor": {
"@type": "Organization",
"name": "Pillsbury House + Theatre" // the company
}
}
}
One detail to watch when you split text like this: one reviewer's title is "Associate Director, Marketing, Optum Everycare", which has two commas. Split at the last comma, not the first, or the company becomes "Marketing, Optum Everycare". These testimonials still will not earn stars, since they are self-serving reviews, but now the details are accurate.
What about other schema types?
A quick reference for the types you will run into most often. Each follows the same rule: describe what is visible.
| Type | Use it for | Search feature today |
|---|---|---|
| Product | A specific item for sale, with price and availability | Yes: price, availability, and ratings in results and shopping surfaces |
| VideoObject | A video embedded on the page | Yes: video results and key moments |
| JobPosting | An open position | Yes: Google's job listings |
| Course | A course with a provider and description | Limited and changing; check Google's current docs |
| HowTo | Step-by-step instructions | No. Google stopped showing HowTo rich results in 2023. The markup is harmless. |
| SoftwareApplication | An app or software product | Yes, with ratings and price |
Google's list of supported features changes every year. When you are deciding whether a type is worth the effort, check Google's structured data search gallery for the current list, and treat anything else as clarity rather than a search feature.
How should a creative team work with schema?
The teams that keep schema healthy share one habit: schema is generated from the same content as the page, not written separately. If the event date in the hero and the event date in the schema both come from one field in the CMS, they cannot disagree.
A practical split of responsibilities:
- Developers build the templates that turn CMS fields into schema, once per page type, and run the validator.
- Content and creative teams fill those fields well: real photos, real author bios, exact dates, honest answers. They also review new templates with the checklists in this series.
- Whoever owns the site spot-checks a few pages every quarter, because schema breaks silently.
That last point is the one most teams skip. My own site had duplicates and mismatches that no visitor would ever notice (they are fixed now, and Part 1 shows how I found them), and a Semrush site audit flagged five invalid structured data items before I went looking. If you only check schema when a page launches, you will not catch what drifts later.
The content schema review
| Question | If the answer is no |
|---|---|
| Is every fact in each block visible on the page? | Remove it from the schema or add it to the page. |
| Do FAQ questions and answers match the page word for word? | Generate the schema from the same content as the visible FAQ. |
| Is each Person described once, with one @id and one job title? | Consolidate on the bio page and point to it. |
| Do event dates include times and time zones, and is eventStatus current? | Update the template and the content. |
| Does each recipe have a real photo, correct times, and full ingredients and steps? | Fix the recipe card fields the plugin reads from. |
| Do ratings reflect reviews shown on the page, about the specific thing reviewed? | Remove ratings that are not visible or that are about your own business. |
| Does every block pass the Rich Results Test? | Fix errors first; a broken block is ignored entirely. |
That completes the series. If you have read all three parts, you can now open almost any page, find its schema, and tell whether it is complete, connected, and honest, which is more than most website teams can say. Start with part 1 if you want the foundations again, or read schema markup for AI search for how all of this relates to AI answers.
FAQ, Person, Event, and Recipe schema FAQ
Does FAQ schema still work?
FAQ markup is still valid, but since 2023 Google has shown FAQ rich results mainly for well-known government and health websites. For most sites it no longer adds expandable questions in search, but it still states your answers clearly for search engines and AI systems when the FAQ is genuinely on the page.
What is the most important rule for schema markup?
The schema must describe content that a visitor can see on the page. Questions, ratings, prices, dates, and other details in the schema should match the visible page exactly.
What does PT20M mean in recipe schema?
It is an ISO 8601 duration. PT means a period of time, followed by hours (H) and minutes (M). PT20M is 20 minutes, PT1H is one hour, and PT1H30M is an hour and a half.
Can I add review stars to my business with schema?
Not for reviews you publish about your own business. Google does not show review stars for self-serving reviews in LocalBusiness or Organization markup. Ratings can appear for specific things such as products, recipes, or courses when the reviews are shown on the page.
What does Event schema need?
At minimum a name, a start date with a time and time zone, and a location, either a physical Place with an address or an online location with a URL. Event status, attendance mode, end date, image, organizer, and ticket offers are strongly recommended.
Explore related Double Atari resources: schema explained, part 1, part 2, schema markup for AI search, and content strategy.