How long should a title and description be?

Short enough to display without truncation and specific enough to distinguish the page. Measure what the search result shows rather than trusting a character count.

Does duplicate metadata hurt rankings?

It removes the main reason a specific page would be chosen over another. The practical loss is click-through, and the deeper risk is confusion about which page should rank.

Can a static check validate structured data?

It can check syntax and presence. Whether the claims match the page is a comparison against rendered content, which a person or a crawler does.

Should every page have structured data?

No. Use the types the page genuinely represents. Adding markup for content that is not there creates a mismatch that can suppress rich results.

How often should metadata be audited?

Before every launch, after every redesign and on a quarterly sample. The most common regression is a template change that drops a field.

Web

A metadata audit is about what a stranger sees before they click.

The title, the description, the social image and the canonical URL decide whether a page is ever read. They are also the fields most likely to be generated once and never revisited.

Run a free check Request a human review

The fields worth auditing

A metadata audit is a small list, applied consistently. Most of the value comes from catching duplication and drift rather than from clever optimisation.

Title: unique per page, specific about the page rather than the brand, inside the length a search result displays.
Description: written to be read, not to repeat the title, and honest about what the page delivers.
Canonical: absolute, self-referencing where appropriate, and consistent with the internal linking.
Social tags: og:title, og:description, og:image with a real image at the right ratio, and a Twitter card type.
Structured data: the types that apply to the page, valid against the schema, and matching the visible content.
Indexing directives: robots meta and sitemap inclusion consistent with the page's purpose.

Duplication is the default failure

Generated pages inherit templates, and templates inherit defaults. The result is a site where forty pages share one description and the same social image, which is invisible on the site and obvious in a search result.

Check for identical titles, identical descriptions and missing values across the whole set rather than page by page. That comparison is where the findings are, and it needs a list of pages rather than a single document.

The same applies to structured data: a template that emits an organisation block on every page produces dozens of near-identical claims, some of them contradictory.

Structured data has to match the page

Markup is a claim about content, and a claim that contradicts the page is worse than no markup. A FAQ block for questions that are not on the page, a rating with no reviews behind it, a price that differs from the visible one.

Validate the syntax and then read the claims. Syntax errors are found by tools; contradictions are found by comparing the block against the rendered page.

Keep it to types that describe the page. A page usually needs one or two, not six.

Titles and descriptions that survive rewriting

Generated titles drift toward the generic: a product name plus a benefit phrase that could describe anything in the category. The page then competes on nothing.

Write the title from the question the page answers, and the description from the promise it keeps. A description that overstates is clicked once and abandoned, which is a measurable loss rather than a style preference.

Where a page is generated in bulk, generate the metadata from the page's own specifics rather than from the template. The heading and the first sentence are usually the best source.

The audit cadence

Run a full pass before a launch and a sampled pass quarterly. Templates change, CMS fields get added, and the social image stops matching the product.

Also audit after any redesign, because metadata is the first thing a new template drops. A missing canonical after a migration is a cheaper finding than a month of duplicate indexing.

Keep the audit output: a table of page, title, description, canonical and findings. It is what makes the next audit a comparison rather than a discovery exercise.

What a document check covers, and what it cannot

Pasting a page's markup into the console surfaces the mechanical issues: a missing title element, absent lang or viewport declarations, placeholder links, images without alt text, and heading structure problems.

It cannot fetch the rest of the site, so cross-page duplication and canonical consistency need a crawl or a list you assemble yourself.

Nor can it judge whether a description is any good. That is an editorial call, and for anything carrying regulated claims it ends in a named approval.

What this covers

Unique titles
No two pages sharing a title, and none defaulting to the brand alone.
Descriptions written
Page-specific copy rather than a template string repeated site-wide.
Absolute canonicals
Consistent with internal links and correct after any migration.
Social preview
Real image at the right ratio, with matching title and description.
Structured data checked
Valid, and every claim matched against the visible page.

Read the method

Questions this page usually raises

How long should a title and description be?: Short enough to display without truncation and specific enough to distinguish the page. Measure what the search result shows rather than trusting a character count.
Does duplicate metadata hurt rankings?: It removes the main reason a specific page would be chosen over another. The practical loss is click-through, and the deeper risk is confusion about which page should rank.
Can a static check validate structured data?: It can check syntax and presence. Whether the claims match the page is a comparison against rendered content, which a person or a crawler does.
Should every page have structured data?: No. Use the types the page genuinely represents. Adding markup for content that is not there creates a mismatch that can suppress rich results.
How often should metadata be audited?: Before every launch, after every redesign and on a quarterly sample. The most common regression is a template change that drops a field.

Run the checks, then decide who has to sign it.

The automated pass is free, needs no account and uploads nothing. Escalate to a named reviewer only where being wrong is expensive.

Run a check Request a specialist