Website Metadata and Schema: How MD3 Checks That Google and AI Tools Can Index Your Site
D
Dariel Corchado
MD3 Digital Solutions
Short answer: website metadata and schema are the information your site hands to Google and AI tools so they know what your business is, what it offers and which page to show. At MD3 Digital Solutions we check them page by page, validate the schema with the same tools Google uses and fix whatever is missing or wrong. Here’s how we do it.
A site can look great and still be close to invisible. We see it all the time: a well-designed services page that, once you open the code, has no meta description, shares its title with twelve other pages and never says which city the business works in. Google crawls it but doesn’t quite know what to do with it. Neither does ChatGPT.
What website metadata is, and who reads it
Metadata is a set of tags in each page’s code that visitors almost never see. Search engines read it. Social networks read it when someone shares a link. And more and more, AI assistants read it when they look things up on the web before answering.
Schema (also called structured data) goes a step further. If the metadata says “this page is about roof repair,” the schema says “this is a local business, here’s its name, its address, its hours and its phone number.” And it says it in a format a machine can read without guessing.
What we check on every page
We start with the basics, because that’s where we find the most mistakes:
Title tag. Unique for every page, around 60 characters, with the phrase your customer would actually type into Google. A title like “Home” wastes the most valuable spot in the search result.
Meta description. It doesn’t raise your ranking directly, but it affects whether people click. If it’s missing, Google picks a snippet from the page, and not always the best one.
Canonical tag. Tells Google which version of a page is the official one when several URLs show the same content.
Indexing directives. We look for noindex tags left in by mistake. The classic WordPress case: the “Discourage search engines from indexing this site” box was still checked after the site moved from staging to production.
Open Graph tags. They control the title, image and description that show up when someone shares the link on Facebook, LinkedIn or WhatsApp.
Hreflang. If your site is in English and Spanish, these tags tell Google which version to show each person. Get them wrong and your Spanish page can turn up in English searches.
Headings and alt text. One clear H1 per page, and alt text that actually describes each image.
Schema: your business’s spec sheet
We use JSON-LD, the format Google recommends. It’s a block of code that changes nothing visitors see. Here’s a simple example for a local business (the details are made up):
Depending on the site, we work with Organization or LocalBusiness, WebSite, Article for the blog, BreadcrumbList for navigation, Service for service pages and Product for online stores.
One rule we don’t bend: the schema has to say the same thing as the page. If the code advertises hours or reviews that visitors can’t see, Google may ignore that data or, in serious cases, take manual action against the site.
A straight answer about FAQ schema (FAQPage), too. Since 2023, Google only shows that expandable FAQ in search results for well-known government and health sites. We still use it when a page has real questions, because it helps search engines and AI understand the content. What we won’t promise you is the dropdown.
The tools we use to validate schema
No single tool checks everything, so we combine several:
Google’s Rich Results Test (search.google.com/test/rich-results). Shows whether a page is eligible for rich results in Google and flags missing required fields. It takes a URL or a code snippet, so we also use it before anything goes live.
Schema Markup Validator (validator.schema.org). Maintained by Schema.org, it checks the whole vocabulary, not just the part Google uses for rich results.
Google Search Console. The URL Inspection tool shows which structured data Google found in its saved copy of the page, and the enhancement reports warn you when new errors appear.
Screaming Frog SEO Spider. Crawls the whole site the way a search engine would. One report shows us duplicate titles, empty meta descriptions, canonicals, hreflang, and schema errors across hundreds of pages.
Lighthouse and PageSpeed Insights. Their SEO audit covers the basics: the page can be indexed, it has a meta description, and its canonical and hreflang are valid.
Facebook’s Sharing Debugger and LinkedIn’s Post Inspector. They show how a link will look when shared and let us refresh the cached version after fixing the Open Graph tags.
Bing Webmaster Tools. People forget about Bing, but its index feeds Microsoft Copilot and other search assistants. We make sure your site is indexed there too.
The first two answer different questions. Schema can pass Google’s test and still get warnings in the Schema.org validator, or the other way around. That’s why every page goes through both.
How the review works, step by step
We crawl the entire site with Screaming Frog and export titles, descriptions, canonicals, hreflang and structured data.
We pick one page from each template (home, service, blog post, contact) and review its code by hand. If a template has an error, every page built on it has the same error.
We run those pages through the Rich Results Test and the Schema.org validator.
We review the robots.txt file and the XML sitemap.
We fix things in WordPress, usually with Rank Math, and write custom schema where the plugin falls short.
We validate again, ask Google to recrawl the main pages from Search Console and check the reports a few weeks later.
At the end you get a clear list: what we found, what we fixed and what’s still on your side, like filling out your Google Business Profile.
What about AI tools?
ChatGPT, Claude, Perplexity and Google’s AI Overviews need the same thing a search engine does: they have to be able to get into your site and understand what they find. That’s why we go over robots.txt closely. Some sites block every AI bot without knowing it, sometimes because of a rule a security plugin added.
One distinction is worth knowing. OpenAI uses GPTBot to train its models and OAI-SearchBot for ChatGPT search. You can block the first and allow the second, so your site can be cited in ChatGPT without your content being used for training. Anthropic (Claude) and Perplexity have their own crawlers too, and we make sure robots.txt treats each one the way you want.
What about llms.txt? It’s a recent proposal, and no major search engine has confirmed it uses it. We’ll add one if you want, but we don’t put it ahead of solid metadata and schema.
If you’re not sure whether your site has this metadata, it’s probably missing some of it. That’s normal. Almost nobody reads the source code of their own website.
At MD3 Digital Solutions we review your website metadata and schema, validate everything with the tools above and fix what’s needed in WordPress. We explain every change in plain English.
Want to know what Google and AI tools see when they visit your site?Contact us and we’ll put together a review for you.
Agencies are selling "GEO" as a secret trick. Google's own guide says there are no extra technical requirements: appearing in AI Overviews is good SEO. Here's what to do instead.
Google's Search Console now has a Generative AI Performance report. Here's how to find it, what it measures (and what it doesn't), and how to read it without jargon.
Since late October 2026, Google has been showing the AI Overview for many brand searches where the knowledge panel used to be. Here's what changed and how to check what AI says about your company.