New docs. Ask anything via the floating widget in the bottom-right.
DiagnoseSEO health
Diagnose

SEO health

On-page and technical issues on your own site, read from the crawl AnswerTrace already runs during a scan. No extra crawl, no separate tool to install.

What this is for

Most of what stops AI engines citing you also stops Google ranking you. A page with no title, no schema, or 80 words of content is invisible to both.

So this section is not a second product bolted on. It is the same crawl, read through a different lens. AI wants to understand your page. Google wants to rank it. Both need the page to be readable first.

Available on Starter and above.

The nine categories

Every finding falls into one of nine, always shown in the same order so you can scan for what changed:

CategoryWhat triggers it
Title tagsMissing, outside 30 to 60 characters, or duplicated across pages
Meta descriptionsMissing, or outside 70 to 160 characters
H1 headingsNo H1 on the page
Canonical tagsThe canonical points at a different domain
Product / Offer schemaA money page with no structured data at all
Thin contentAn indexable page under 250 words
Broken internal linksA link pointing at a page of yours that does not load
Noindex on key pagesA page that should rank is telling search engines to skip it
CSR-only key pagesThe content only exists after JavaScript runs

Each category shows its worst severity, how many pages are affected, a sample of the URLs, and what to do about it.

Reading the severities

Four levels, and they mean different things:

  • Fail. Broken. A missing title, or a link into a dead page.
  • Warn. Working but costing you. A title outside the length band, a page under 250 words.
  • Info. Worth knowing, not worth a sprint. A meta description slightly off length.
  • Ok. Checked and clean.

Two deliberate choices in how these are counted.

A page we could not fetch is not counted as an issue. If the crawl could not reach it, we do not know whether it has a title, so calling it a failure would be a guess.

Legal pages are exempt from thin content. A privacy policy is short by nature and is not a ranking target. Counting it would produce a warning nobody should act on.

Quick wins

Three of the nine can be drafted for you: title tags, meta descriptions, and Product / Offer schema. The section counts these separately as quick wins, because they need writing rather than engineering.

The other six need someone to change the site. Fixing a canonical, removing a noindex, or prerendering a page is a code change, not a copy change.

That split is the useful read. Quick wins are this afternoon. The rest go to whoever owns the site.

Where you rank on Google

Alongside the on-page findings, a panel shows your Google position for the same buyer questions AnswerTrace tracks in AI answers.

This is the bridge, and it is the reason the section exists in an AI visibility product. Four states are worth naming:

  • Ranking on Google and named in AI. Working. Leave it alone.
  • Ranking on Google, not named in AI. The interesting one. Google can read the page and AI is not citing it. Usually a structure or citation problem, not a content one.
  • Named in AI, not ranking on Google. Rarer. Something off your site is carrying you.
  • Neither. You have no presence on that question at all.

The second row is where most of the value is. It is a gap you can only see by holding both reads together.

What this is not

This is not an SEO suite. There is no rank tracker, no backlink index, no keyword research tool, no site-wide technical crawl of every URL.

If you use Semrush or Ahrefs, keep them. This section exists so the AI visibility work and the SEO work stop being two separate conversations, not so you can cancel a subscription.