What is broken, and on which pages?
A site audit from a crawl of your pages, read-only Search Console access and speed data from real visitors.A site audit of your own pages.
80 checks on your own pages, 27 of them E-E-A-T signals. Free for one site.
LCP2140ms
CLS0.04
INP236ms
We read your pages, then pin each problem.
<title>Northwind pricing: product analytics plans for small SaaS teams and startups</title><h1>Pricing that grows with your events
</h1>
<p class="lede">Product analytics for small SaaS teams.</p>
<section id="plans"><h2>Free</h2>
<data>$0</data>
<li>10k events</li>
<h2>Team</h2>
<data>$49</data>
<li>1m events</li>
<h2>Scale</h2>
<data>$149</data>
<li>10m events</li>
</section><a href="https://northwind.example/
docs/legacy-sdk">
Read the legacy SDK notes</a> Broken link/blog/event-naming<title>Event naming that lasts: a guide for product teams</title><script type="application/ld+json">
{ "@type": "BlogPosting",
"headline": "Event naming that lasts",
"publisher": { "name": "Northwind" },"author":}
</script> Missing Author Markup<p>Old names can live on for a while. Our
<a href="https://northwind.example/docs/legacy-sdk">legacy SDK reference</a>
lists the aliases.</p> Broken link same dead URL<aside>Ship events your team can read next year.</aside>
</body>Ranked by severity,
then by reach
4 of the 17 issue types. Warning 3, Info 14.
Pick a pin.
See what to do.
Its pages, why it matters, and the steps.
Broken Links
“Internal link to https://northwind.example/docs/legacy-sdk returned 404”
- /pricing first seen 20 days ago
- /docs/api first seen 20 days ago
- /blog/event-naming first seen 6 days ago
Why it matters
A confirmed dead destination can interrupt a useful journey. A timeout or transient server failure does not establish a permanently broken link.
Steps
Confirm the applicability and required context before preparing any change.
Applies after a GET-capable probe confirms an unintended dead destination. Prioritize internal/key-journey links and repeated failures; transient 5xx remains investigation. Correct the href, restore the target, or remove the link.
Verify the result against the conditions below.
Preserve
Preserve complete authored meaning, accurate claims, accessibility, security, and intentional crawler/canonical policy.
Avoid
Redirect every dead URL home, replace with an unrelated destination, remove useful anchor context, or treat a timeout as a 404.
Leave it alone when
Keep the current implementation when the applicability conditions are not met or the existing implementation already serves its intended purpose.
Verify
Link reaches the intended relevant resource with expected status/content; key journey and anchor meaning still work.
Missing Author Markup
“No author attribution found in JSON-LD, meta tags, or HTML markup”
- /blog/event-naming first seen 44 days ago
- and 25 more pages
Why it matters
Truthful attribution can help readers understand responsibility for authored content; optional markup must not invent an author.
Steps
Confirm the applicability and required context before preparing any change.
Applies only to genuinely authored editorial/how-to content with a known responsible person or organization. Advisory: expose truthful visible attribution and optionally matching Article markup.
Verify the result against the conditions below.
Preserve
Preserve complete authored meaning, accurate claims, accessibility, security, and intentional crawler/canonical policy.
Avoid
Invent a person, use a fake expert, add hidden/schema-only attribution, or conflate publisher with author.
Leave it alone when
Do not add an author to navigation, utility, or collection pages that are not genuinely authored editorial content.
Verify
Visible byline and structured data, if used, identify the real responsible author consistently.
Title may be shortened in search results
“Title tag contains 76 characters.”
- /pricing first seen 31 days ago
- and 13 more pages
Why it matters
Length alone does not establish an editorial defect; clear, accurate metadata can be retained.
Steps
Confirm the applicability and required context before preparing any change.
Keep it if it is clear and accurate. Length-only suggestion. Keep if clear; edit only for a separately identified problem such as repetition or misleading wording.
Verify the result against the conditions below.
Preserve
Preserve complete authored meaning, meaningful qualifiers, page distinctiveness, and accurate branding composed once.
Avoid
Cut at 60/70 characters, cut mid-phrase, or bulk rewrite. Do not mechanically cut, pad, bulk rewrite, or introduce shared truncation rules. Character and pixel counts are observations, not mandatory editing ceilings.
Leave it alone when
Keep the complete metadata when it is clear, accurate, relevant, and distinctive, regardless of character count or preview width.
Verify
Complete intended wording survives rendering; page remains distinctive; brand composed once.
Orphan Pages
“Page has no inbound links from other crawled pages”
- /docs/feature-flags first seen 6 days ago
- and 8 more pages
Why it matters
An important page may be hard to discover without contextual links, but a sampled link graph cannot prove that it is orphaned.
Steps
Confirm the applicability and required context before preparing any change.
Investigate only for an important index-oriented page after a complete link graph; sampled crawl absence is weak evidence. Add a contextual internal link or intentionally exclude/retire it.
Verify the result against the conditions below.
Preserve
Preserve complete authored meaning, accurate claims, accessibility, security, and intentional crawler/canonical policy.
Avoid
Add site-wide footer links, force every page into navigation, or expose private/utility URLs.
Leave it alone when
Keep intentionally isolated, private, or utility pages out of navigation; investigate incomplete graph evidence first.
Verify
Important page is reachable from relevant navigation/content; intentional isolates are recorded N/A.
Speed, as real visitors felt it
LCP and CLS are in the good range. INP is the one to watch.
LCP
CLS
INP
of 27 trust signals found
Each bar is one signal. The red ones are the 8 we did not find.
Experience 4 of 6
Expertise 5 of 6
Authoritativeness 2 of 6
Trustworthiness 8 of 9
Not found
- Has case studies
- Has original research
- Has credentials displayed
- Has guest contributions
- Has media coverage
- Has industry partnerships
- Has testimonials
- Has referring domain breadth
4 resolved in
the last run
Resolved lists what the next run stopped seeing. A pin stays until the problem is gone.
Re-run audit is free once per hour.
Start free/docs/webhooks
Resolved
/changelog
Resolved
/blog/funnel-basics
Resolved
/docs/getting-started
Resolved
/pricing
Still open
Example data
Where the line falls
| Plan | What you get |
|---|---|
| Free Starts here | The full audit — one site on Autopilot: up to 5,000 pages kept fresh every 30 days. Every check, the issue queue, and 30 days of score history. |
| Starter · $9 | The wider cut — up to 10,000 pages kept fresh every 14 days per site on Autopilot, and no site cap. |
| Pro · $49 | The deepest record — up to 25,000 pages kept fresh every 7 days per site on Autopilot, and full score history. |
Frequently asked questions
What does one TrustGrowth audit actually check?
Eighty distinct checks against your real pages: meta and headings, links in and out, schema, Core Web Vitals, GEO readiness, and the E-E-A-T signal set. Every finding is filed with the pages it affects and how long it has been open, so "fix this" always has an address.
Who calculates the TrustGrowth Score, and is it a Google number?
It is ours, and the product labels it as ours. Google numbers on your account are the Search Console ones — clicks, impressions, CTR, position — and they are marked separately. The score is one number per day per site, drawn against a dated baseline.
Will fixing an issue raise my score by a set amount?
No, and the product will not tell you it will. Signals interact, so the honest answer to "what did that gain me" is to run the next audit and read it. The audit measures evidence on your site; it does not promise rankings.
What happens when a check cannot run?
It is reported as unmeasured — never as a pass, never as a failure. A page the crawl could not read, a signal with no evidence either way, a check whose input is missing: all of them stay pending rather than being resolved into a number that looks confident.
What are the 27 E-E-A-T signals, and do they all count toward the score?
They are checked across four groups — experience, expertise, authoritativeness, trustworthiness — and each present-or-absent result carries the evidence and the date that decided it. Five of them are displayed but do not feed the score, and the panel badges them "checked · not scored" rather than hiding the seam.
Does the audit re-crawl my whole site every time?
No. Autopilot revisits a bounded set of pages on your plan's freshness window, so the entitlement is two things — how many pages, and how fresh — not a single cadence word. The plan rows above this FAQ render the current figures.
Is the audit free?
Yes, for one site, with the full check list, the whole issue queue, and score history. Paid plans widen the page budget and shorten the freshness window; every check stays on Free.
What this does not say
The audit doesn't promise rankings, and it fixes nothing by itself.
See what is broken.
Free for one site. Sign in with Google to start.