Help Google, AI and customers understand your pages: titles, sharing previews and structured data

Six examples covering services, locations, products, articles, brands and MathFun explain page titles, Open Graph, URLs and structured data for search, AI search and sharing, with checks owners can request.

Customers may see your page in Google, read about your business in an AI answer, or receive a link from a friend before they open your website. Page information helps those systems identify whose website this is, what a page covers, which image represents it and which facts can be checked.

Metadata means information describing a page. Open Graph, or OG, is a set of page tags used by sharing platforms. Structured data uses agreed fields to describe things such as businesses, products and articles. Owners do not need to write the code themselves, but should know what to prepare and check.

This guide starts with business situations, then explains settings and code. Brands, addresses, prices, authors and outcomes in the six examples are illustrative. MathFun illustrates settings for a practice page; its current website configuration and business claims have not been verified. Do not publish these examples as your own business details.

Suppose ClearAir repairs residential central air conditioning in Pasadena.

Where the page appears	What the customer wants to know	What the page needs
Google Search	Does this page offer the repair I need?	A specific title, relevant content and an accurate summary
AI search answers	Do they serve Pasadena, repair central AC and accept appointments?	Readable, checkable coverage, equipment types and booking conditions
Chat or social sharing	What does the business my friend recommended do?	A title, summary, image and URL for this service page

An unhelpful preview: “Services — ClearAir”, a generic logo and “Professional team, quality service”.

A useful preview: “Central AC Repair in Pasadena”, a relevant repair image and a summary explaining the customer, area and booking process.

The second gives customers more information to judge relevance and helps someone introduce the link. More clicks or enquiries would need to be measured. A mock preview cannot establish a conversion improvement.

OG primarily supplies sharing information. Google can also consider og:title for search titles and the homepage’s og:site_name for the site name. Google chooses the final display automatically. See Google: Title links and Google: Site names .

2. Understand what each setting does

Setting	Plain-English meaning	What owners should check
`<title>`	The page’s name, an important source for search titles	Specific, accurate and concise
`meta description`	A page summary offered to search systems	Explain content and purpose without stuffing keywords
H1 and body content	What customers can actually read	Answer the need and deliver the title’s promise
`rel="canonical"`	A preferred-URL signal for duplicate or similar pages	Point to the corresponding production page
OG and X Card tags	Information introducing a shared link	Copy, image and destination agree
`WebSite` and `Organization`	The website’s name and operating organization	Brand, URL, organization and official contact details
Page-specific structured data	Fields describing products, locations, articles and other facts	Suitable types, accurate facts and agreement with visible content
Crawling, indexing and access controls	Whether systems can read and use content	Important pages are not accidentally blocked

Google primarily creates search snippets from body content and sometimes uses the meta description. Different queries may produce different snippets. A description does not fix every display. See Google: Snippets and descriptions .

A canonical is a preference signal, not an absolute instruction. og:url does not replace it. Do not canonicalize every distinct service page to the homepage. See Google: Consolidating duplicate URLs .

3. Six examples: prepare facts before choosing tags

Example A: A service page that never explains what or where

Before: The page is called “Services”, with “Professional repairs. Contact us.” as its main explanation.

An improved example:

Page element	Hypothetical content
Page title	Pasadena Central AC Repair \| ClearAir
H1	Central AC repair for Pasadena homes
Description	Central AC troubleshooting and repairs for Pasadena homes. See service coverage, appointment hours and how estimates work.
Body	Equipment repaired, actual service areas, appointment hours and estimate process
Sharing title	Central AC Problems in Pasadena? See Our Repair Services
Sharing image	An image representing the central AC repair service

Google gets a more specific topic, and customers can judge suitability. In an AI service comparison, those body facts also provide information to check. This is content preparation advice, not a claim that a particular tag increases AI recommendations.

A developer can consider Service markup when it accurately describes the content. A type existing in Schema.org does not mean Google offers a dedicated rich result for it. Check Google’s supported structured data features first. Do not disguise an ordinary service page as a rated product.

Check: After opening the page, can customers find the service, coverage and booking details promised by the title?

Example B: A location with conflicting hours and addresses

Suppose Willow Yoga moves. The website body has the new address, but old markup and Google Business Profile retain the old one. Sunday opening hours also disagree.

Confirm the facts, then update the location page, Business Profile and applicable LocalBusiness data together. These fictional studio fields illustrate how information corresponds:

{
  "@context": "https://schema.org",
  "@type": "LocalBusiness",
  "name": "Willow Yoga",
  "url": "https://example.com/studio",
  "telephone": "+1-202-555-0148",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "123 Example Street",
    "addressLocality": "Pasadena",
    "addressRegion": "CA",
    "postalCode": "91101",
    "addressCountry": "US"
  },
  "openingHoursSpecification": [{
    "@type": "OpeningHoursSpecification",
    "dayOfWeek": ["Saturday", "Sunday"],
    "opens": "09:00",
    "closes": "13:00"
  }]
}

This example lists weekend hours only, not a complete schedule. In production, choose the most accurate applicable subtype, add real hours and meet the official requirements. Do not publish a fictional address or invent a storefront for local visibility.

Google supports local business data describing names, addresses, hours and other information, without guaranteeing special display. Website tags do not automatically update Business Profile. See Google: Local business structured data .

Check: Do the website, markup and Business Profile describe the same location and hours? Explain class times separately from business opening hours.

Example C: A product with ambiguous price, size and stock

Suppose a store sells 250g coffee beans. Its title says “Premium Coffee”, the price appears only in an image, and the page says sold out while old data still says in stock.

Display the product name, size, price, currency and availability in text. Explain delivery and returns, then keep product data aligned with those facts.

{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "House Blend Coffee Beans, 250g",
  "image": "https://example.com/images/house-blend-250g.jpg",
  "description": "A 250g bag of whole-bean house blend coffee.",
  "sku": "HB-250",
  "offers": {
    "@type": "Offer",
    "url": "https://example.com/coffee/house-blend-250g",
    "price": "18.00",
    "priceCurrency": "USD",
    "availability": "https://schema.org/OutOfStock"
  }
}

This teaching example does not include every required or recommended field for every merchant feature. Google can use applicable product data to display prices, availability and other details. Pages where customers can buy and editorial review pages have different requirements. See Google: Product structured data .

For an AI query comparing 250g coffee prices and purchase conditions, the page offers explicit facts. Adding Product does not guarantee an AI recommendation. Google’s AI guidance also recommends keeping Merchant Center, Google’s merchant product management platform, up to date. That channel and page markup need separate maintenance. See Google: AI features and your website .

Check: After changing price or stock, are the body, markup and product data channels updated together? Does the sharing image represent this size?

Example D: An article with no clear author or applicability

Suppose ClearAir publishes “Why Is My AC Leaking Water?” The old version has no author, dates or applicable conditions, and contains only generic advice.

Add the author and their actual experience, inspection steps, relevant equipment, sources and circumstances requiring a professional. Show real publication and update dates, then describe the same article in its data:

{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "Why Is My AC Leaking Water?",
  "image": "https://example.com/images/ac-leak-guide.jpg",
  "datePublished": "2026-09-10T09:00:00-07:00",
  "dateModified": "2026-10-05T10:00:00-07:00",
  "author": {
    "@type": "Person",
    "name": "Alex Example",
    "url": "https://example.com/team/alex"
  }
}

Google can use article data to understand the title, author, images and dates. See Google: Article structured data .

For AI search, prepare experienced content that can support an answer. An author tag cannot establish expertise, and changing a date does not make outdated advice current. dateModified should reflect an actual change. Google recommends useful, distinctive content for readers. See Google: Generative AI search guidance .

Check: Does the author link work? Are sources and applicability explained? Do visible dates match the data?

Example E: A homepage with an easily confused brand name

Suppose Northstar is a web studio, but logistics and finance businesses share the name. “Welcome to Northstar” does not explain which business this is.

Explain the business, customers, operating organization and official contact information. Use WebSite for the website’s name and Organization for the organization. They are related but describe different things.

{
  "@context": "https://schema.org",
  "@graph": [
    {
      "@type": "WebSite",
      "@id": "https://example.com/#website",
      "url": "https://example.com/",
      "name": "Northstar Studio",
      "publisher": { "@id": "https://example.com/#organization" }
    },
    {
      "@type": "Organization",
      "@id": "https://example.com/#organization",
      "name": "Northstar Studio",
      "url": "https://example.com/",
      "logo": "https://example.com/images/northstar-logo.png"
    }
  ]
}

@id is a stable address identifying an object in the data. It does not require a separate page displaying JSON. sameAs can add verified official profiles representing the same organization. Not every page mentioning the business is an official identity link.

Google uses homepage WebSite data as an important source for site-name preferences. Organization data can help distinguish organizations. See Google: Site names and Google: Organization data .

Clear, consistent public identity is content preparation advice. No universal tag forces every AI service to identify similarly named sources as your brand.

Check: Do the name, URL, logo and contacts agree? Have links to a different organization with the same name been included accidentally?

Example F: MathFun must deliver the preview’s promise

The MathFun teaching example points to /practice . Its title describes free math practice for Grades 1–5 with answers. Its summary explains the four basic operations and how practice works.

Page element	What this example should provide
Visible content	Supported grades and operations; selecting questions, practising and checking answers
Search information	An accurate page title and description
Sharing information	The practice page’s title, summary and relevant image
Production URL	Canonical and `og:url` individually pointing to this page
Breadcrumb	A real “Home → Math Practice” relationship

BreadcrumbList describes page hierarchy. It is structured data, separate from OG. Google supports breadcrumb display in desktop search results. See Google: Breadcrumb data .

Check: Can customers find the promised grades, practice and answers after clicking? “Free” or “With Answers” must not exist only in metadata.

4. Open Graph: a link’s first impression

OG supplies sharing information in HTML’s <head> . Its core properties are title, type, image and URL. The summary and brand name add context. Image URLs should be publicly accessible. See the Open Graph protocol .

Parameter	What it means to an owner	Common mistake
`og:site_name`	The site’s brand	Retaining an old name
`og:type`	Page type, such as `website` or `article`	Assigning types without considering the content
`og:title`	The title introducing a shared page	Every page says “Home”
`og:description`	One or two sentences on content and purpose	Disagreement with the destination
`og:url`	The shared object’s production address	A service page points to the homepage
`og:image`	An image representing the content	Inaccessible or unrelated image
`og:image:type`	The image file type	Declaration differs from the actual file
`og:image:width` and `height`	Actual pixel dimensions	Declaring dimensions without creating that image
`og:image:alt`	A description of the image content	A keyword list
`og:locale` and `alternate`	Content language and region	Values do not reflect the language
`twitter:card` , `title` , `description` , `image` and `image:alt`	Card type and content offered to X	Outdated differences from OG or the page

og:image:alt does not replace a body image’s alt . A sharing image also does not replace relevant product or article images in the body. A 1200×630 image is a practical starting point, not a universal platform requirement. Dimension tags do not generate or resize images. Platforms may crop, shorten or omit content.

Use a brand introduction for the homepage, a service image for its page, the current product for a product page and a topic image for an article. Keep the subject readable on small screens and away from edges. These layout suggestions should be checked in the actual target platforms.

If an old image persists, check live HTML, image access, duplicate tags and platform caches. Changing a page does not mean platforms immediately fetch it again.

5. A complete example for your website maintainer

This expands MathFun into an HTML page framework. It assumes the image really is a 1200×630 JPEG and the content and paths exist. Verify and replace values before adopting it. The code is not an audit of MathFun’s live page.

<!doctype html>
<html lang="en">
  <head>
    <meta charset="utf-8">
    <meta name="viewport" content="width=device-width, initial-scale=1">
    <title>Free Math Practice for Grades 1–5 with Answers | MathFun</title>
    <meta name="description" content="Practice addition, subtraction, multiplication and division for Grades 1–5. Choose a grade and skill, solve questions and check your answers.">
    <link rel="canonical" href="https://www.mathfun.me/practice">

    <meta property="og:site_name" content="MathFun">
    <meta property="og:type" content="website">
    <meta property="og:title" content="Free Math Practice for Grades 1–5 with Answers">
    <meta property="og:description" content="Choose a grade and skill, practise the four basic operations and check your answers.">
    <meta property="og:url" content="https://www.mathfun.me/practice">
    <meta property="og:image" content="https://www.mathfun.me/og-image.jpg">
    <meta property="og:image:type" content="image/jpeg">
    <meta property="og:image:width" content="1200">
    <meta property="og:image:height" content="630">
    <meta property="og:image:alt" content="MathFun practice screen with grade selection and arithmetic questions">

    <meta name="twitter:card" content="summary_large_image">
    <meta name="twitter:title" content="Free Math Practice for Grades 1–5 with Answers">
    <meta name="twitter:description" content="Choose a grade and skill, practise and check your answers.">
    <meta name="twitter:image" content="https://www.mathfun.me/og-image.jpg">
    <meta name="twitter:image:alt" content="MathFun practice screen with grade selection and arithmetic questions">

    <script type="application/ld+json">
      {
        "@context": "https://schema.org",
        "@type": "BreadcrumbList",
        "itemListElement": [
          {
            "@type": "ListItem",
            "position": 1,
            "name": "Home",
            "item": "https://www.mathfun.me/"
          },
          {
            "@type": "ListItem",
            "position": 2,
            "name": "Math Practice",
            "item": "https://www.mathfun.me/practice"
          }
        ]
      }
    </script>
  </head>
  <body>
    <nav aria-label="Breadcrumb">
      <a href="/">Home</a> / <span>Math Practice</span>
    </nav>
    <h1>Free math practice for Grades 1–5</h1>
    <p>Choose a grade and skill, then solve questions and check your answers.</p>
    <!-- Add the actual working practice tools and supporting information here. -->
  </body>
</html>

The framework still needs working practice features. A complete set of tags cannot establish that the whole page is finished.

For multilingual pages, maintain real translations, matching language URLs and hreflang . It describes language or regional alternatives; og:locale does not replace it. Each language version should normally canonicalize to itself and link reciprocally to its translations. See Google: Localized page versions .

6. What is supported from an AI SEO perspective?

Here, AI SEO means improving discovery, descriptions and citations in AI search. Services use different data sources and access rules. One platform’s guidance does not describe every AI system.

Google AI search still needs useful pages and search foundations

Google says AI Overviews and AI Mode use existing SEO foundations. To qualify for these search appearances, a page needs to be indexed and eligible for a search snippet. Eligibility does not guarantee selection. No special AI Schema is required, and llms.txt is not a Google Search optimization signal. See Google: AI features and your website and Google: Generative AI search guidance .

Suppose a customer asks: “Are there beginner yoga classes on Saturday mornings?”

Page condition	Which checkable facts are available?
“Quality classes”, with complete OG	The audience and time are still unclear
Body lists beginner classes, Saturday at 10am, location, fee and booking conditions	Facts needed to answer the question
Body matches the timetable, but old data says Sunday	Conflicting information to correct first

Practical advice: Answer real questions on the page, then describe those answers accurately. Clear paragraphs, tables, conditions, sources and dates help people check information. They are not a mandatory AI ranking template. Google recommends useful, distinctive content and does not require fixed-size content chunks. See Google: Generative AI search guidance .

ChatGPT search: separate search access from training access

OpenAI explains that OAI-SearchBot is for ChatGPT search and GPTBot is for model training. The settings are independent. If an owner chooses to permit search crawling while declining training crawling, a maintainer can configure:

User-agent: OAI-SearchBot
Allow: /

User-agent: GPTBot
Disallow: /

This robots.txt example must be integrated with existing rules and checked. It cannot remove login, server or CDN blocks. OpenAI says sites opting out of OAI-SearchBot will not appear in ChatGPT search answers, though they can still appear as navigational links. Access permission does not guarantee citations, recommendations or sales. See OpenAI: Crawler documentation .

Record access and recommendations separately

This is a business checking sequence, not a claim that every AI service uses Google’s index. One mention is not a stable recommendation. One absence does not establish a configuration error.

7. How to verify more than the presence of tags

What to check	Evidence to request	What it cannot prove
Page content	Verify services, locations, prices, stock, authors and dates	Complete tags do not establish useful content
Google access and indexing	Search Console URL inspection and issue resolution records	Access does not mean indexing
Structured data	Rich Results Test for supported types; syntax and vocabulary checks for others	Passing does not guarantee special display; an unrecognized type is not necessarily invalid syntax
Sharing previews	The actual platform’s title, image, summary and destination	One preview cannot establish consistency across platforms
Languages	Accessible versions, complete translations and matching language links	Translating only menus is not a full translation
AI search	Query, date, platform, answer and cited sources	One mention cannot establish rankings or causation
Business outcomes	Relevant visits, actual enquiries, sales and measurement periods	More traffic does not necessarily mean more sales

Start with Google’s Rich Results Test , Search Console and the Schema.org validator . They check different things. There is no single full score covering search, AI and sharing outcomes.

When prices, stock, hours, brand names or articles change, check the body, markup and external business details together. Verify the page, sharing previews and search performance separately, allowing systems time to process updates.

8. What can an owner do today?

Select the homepage, a main service or product page and an important article.
Confirm each page’s title, summary, production URL and relevant image.
Check the facts customers need; add missing coverage, conditions and contacts.
Ask the maintainer to choose suitable data types and check access, duplicate tags, languages and consistency.
Save the verification results, then observe actual search appearances, citations, enquiries and sales.

Owners confirm facts, copy and images. Maintainers implement settings and check access. Both verify actual displays. Keep these settings aligned with changes to the business.