SEO metadata on your website – from title tags and meta descriptions to schema, canonical and hreflang

Definition

Metadata is defined as "data about data" and refers to information about a web page that is not visible in the content but helps search engines, social platforms and AI understand what the page is about. Title tags, meta descriptions, schema markup, canonical and hreflang are all SEO metadata.

What is metadata on a web page?

Metadata is data about data – information that describes the content of a page without itself being visible to the visitor. On a web page, metadata lives mainly in the page's head section and in structured signals such as schema markup. Search engines, social platforms and language models read metadata to decide what the page is about, who is behind it, how it should appear in search results and whether it should be indexed at all.

Two kinds are worth keeping apart: metadata that affects how a page is displayed (for example the title tag and meta description in the SERP) and metadata that helps machines understand it (schema markup, canonical, hreflang and robots directives).

What types of SEO metadata are there on a web page?

In SEO, metadata is usually divided up like this:

• Descriptive metadata – the title tag and meta description. Shown directly in the search results, they affect both CTR and how Google interprets the page's topic.
• Structured data – schema markup (usually in JSON-LD format) that describes entities, FAQs, products, organisations and articles in a format search engines understand directly.
• Social metadata – Open Graph (Facebook, LinkedIn) and Twitter Cards (the technical protocol still has that name, even though the platform is now called X), which control how the page is presented when it is shared.
• Technical directives – robots meta, canonical, hreflang, viewport, charset, the lang attribute on the html element and the favicon. They govern indexing, duplicate handling, language versions, rendering and how the page is identified in browser tabs and bookmarks.
• Content metadata – alt text on images, figcaption and image titles. Technically HTML attributes rather than tags in the head, but in practice they work as metadata: they describe images for search engines, AI models and screen readers.
• Website-level resource metadata – the sitemap and robots.txt. They tell search engines which pages exist and how they may be crawled.

A note on meta keywords – Google has ignored the field since 2009 and it has no place in a modern SEO strategy. Do not spend any time on it.

Does alt text count as metadata?

It is a grey area. Strictly speaking, alt is an HTML attribute on the image element, not a meta tag in the head section. But semantically, alt text is data that describes an image – in other words, data about data. Search engines use alt to understand what the image shows, AI models use it when they interpret the page, and screen readers read it aloud to visually impaired visitors. That is why we include alt text, figcaption and image titles in the broader definition of metadata, even though they do not live in the head.

Practical guidelines: write descriptive alt text that actually describes the image in its context, avoid keyword stuffing, and leave alt empty (alt="") on purely decorative images so screen readers skip them.

Why is metadata important for SEO?

Metadata is the first point of contact between your page and a search engine or AI. Before an algorithm assesses the content, it uses metadata to categorise the page, decide whether it should be indexed, which language version suits the visitor and how it should be presented in the search results.

In practice, metadata affects:

• Indexing – robots meta and canonical decide whether the page is added to the search index and which version counts as the original.
• Click-through rate – the title tag and meta description are all you have to persuade someone to click in the SERP.
• Rich results – correct schema markup can give you star ratings, product prices and other enhanced search results.
• International visibility – hreflang tells search engines which language version to show in which country.
• Authority signals – Organization schema with sameAs links to verified profiles strengthens the website as a recognised entity in Google's Knowledge Graph.

How do AI and answer engines use metadata?

Language models and answer engines such as ChatGPT, Perplexity, Gemini and Google AI Overviews do not build their answers on natural language alone – they also weigh machine-readable signals. Here, metadata has taken on a new and weightier role:

• Schema markup as entity evidence – when an AI confirms facts about a company, a product or an author, it looks for structured data that backs up the claim. If it is missing, there is a high risk that the model leaves the source out altogether.
• Title and meta as a summary signal – AI often uses the title tag as a first signal of what the page is about before it reads the body text.
• Canonical and hreflang – decide which version of a page the AI should cite and which language it should prefer for a given user.
• Author and Organization schema – strengthen the E-E-A-T signals AI weighs when it assesses the source.
• Alt text and image descriptions – multimodal AI models read both the image and the alt text. Good alt text helps the model connect the image to its context and can decide whether the image is cited in AI-generated answers.

If an AI lacks structured information about your product or service, it tends to answer "no" rather than "I don't know" – and that costs you business directly.

Visible versus invisible metadata – what is the difference?

Most people think of metadata as something hidden in the code, but some of it is visible – just in different places:

• Visible in the search results – the title tag (also shown in the browser tab) and the meta description. These must be written for people: short, enticing and keyword-aware.
• Visible when shared – the Open Graph image, OG title and OG description. Not visible on the page itself, but they appear when the link is shared on Facebook, LinkedIn, Slack and the like.
• Visible in the browser – the favicon (in the tab and bookmarks) and the lang attribute (which affects, among other things, how screen readers pronounce the text).
• Invisible – schema markup, canonical, hreflang, robots directives, viewport, charset and alt text. They affect machines and assistive technology without being directly visible to the visitor.

Visible metadata has to work as communication. Invisible metadata has to be technically correct – a wrong hreflang or an incorrect canonical can wipe out a page's visibility entirely without a human visitor noticing anything.

How do you check the metadata on your website?

Checking it is quick – you do not need advanced tools for a basic review:

• View page source (Ctrl+U / Cmd+Option+U) in the browser and scroll through the head section. Title, meta description, Open Graph and canonical are visible straight away.
• Google Search Console – the Pages report shows which URLs are indexed, which are blocked by robots and which have canonical problems.
• Google Rich Results Test – validates schema markup and shows which rich results the page is eligible for.
• Schema Markup Validator (schema.org) – catches syntax errors in JSON-LD.
• Screaming Frog or a similar crawler – gives a website-wide overview of title lengths, duplicates, canonical chains, missing meta descriptions and images without alt text.

Common metadata mistakes

• Identical title tags and meta descriptions on several pages – hurts both CTR and Google's understanding of the website's structure.
• Title tags that exceed Google's SERP width – the rule of thumb is around 60 characters, but the limit is really pixel-based (about 580–600 px). Long words with wide letters get cut off sooner than short ones. Test the titles of critical pages in a SERP simulator.
• Meta descriptions over ~155 characters – Google truncates them or rewrites them completely.
• Incorrect canonical – points to the wrong URL or creates canonical chains that confuse indexing.
• Missing or incorrect hreflang – the wrong language version shown in the wrong country, or the set of attributes left out altogether.
• Schema markup that does not match the visible content – can be interpreted as spam by Google and lead to a manual action.
• A noindex robots meta left over from the staging environment – one of the most common reasons a new website does not show up in Google at all.
• Empty or meaningless alt text on images that carry information – missed ranking opportunities in image search, and a website that is inaccessible to screen reader users.
• A missing lang attribute on the html element – screen readers pronounce the text in the wrong language (Swedish text read with English pronunciation, for example), and Google can misinterpret the language version.

How do you prioritise the metadata work?

Metadata is not a one-off job. A reasonable order of priority for a typical business website:

• Basics (must be there) – a unique title tag and meta description on every page, a correct canonical, robots.txt, a sitemap, the lang attribute and a favicon.
• SEO foundation – Organization and WebSite schema at root level, BreadcrumbList on subpages, alt text on all images that carry information.
• Content-specific – Article, Product, FAQ or Service schema on relevant pages.
• International – hreflang if the website has several language or market versions.
• Social – Open Graph tags and a sharing image at 1200x630.