LLMention
← All 40 checksMetadata & Discoverability

Document language is declared

One of 40 checks the scanner runs. This one is worth 1 point inside the Metadata & Discoverability dimension, which carries 7% of the total score.

Browse sections

What does this check look at?

The <html> element carries a lang attribute.

The attribute's presence is tested, not whether its value is a language anyone speaks, so `lang="english"`, `lang="EN"` and `lang="zz"` all pass. Reading the value for a region subtag is the separate lang-region check, which is why the two can disagree about the same attribute.

How do I fix it?

  1. Declare the language so the page can be retrieved in the right market.

Why is it worth 1 point?

Points are relative weights inside a dimension, and a dimension's weight decides how much of the total it can move. The reasoning for this dimension's weight is published rather than asserted:

DimensionMetadata & Discoverability — 7% of the total score
Points for this check1 of 7 in this dimension
Worth at most1.0 of the 100 points

Moderate, and cheap to satisfy.

Which dimension does this belong to?

Metadata & Discoverability — 7% of the total score, decided by 4 published rules.

Title, description and canonical control what a search or answer surface can show about the page.

None of the five guides covers this rule, which is why this page shows the fix in full rather than pointing at one.

What else is measured in this dimension?

How do I see my own result?

Run a scan on your own domain. The report lists every failed check with the evidence that produced it, so you can tell a real failure from a check that does not apply.

Run a free GEO audit on your own site →

The full method, including the dimensions that are deliberately weighted low and the parts of a picture a single-URL scan cannot see, is on the methodology page.

Questions about this rule

What does this check look at?

The <html> element carries a lang attribute.

Evidence and sources

GEO can boost visibility by up to 40% overall in generative engine responses, and in the paper's Table 1 the best of the tested methods improve on the no-optimization baseline by 41% and 28%. — Generative Engine Optimization, KDD 2024

This rule is applied by the scanner and published in the rule set it is read from, so the condition on this page is the condition the scan tests rather than a summary of it.

Primary sources