☝ One red row in the crawler card - GPTBot blocked - and the criticals sorted above everything else. That's your week's technical to-do, already prioritized.
Fix the website
Site Health sweeps up to 60 of your money pages every week for the technical problems that actually move rankings: HTTPS and redirect sanity, robots.txt - including whether you're blocking the AI crawlers that now send customers - noindex accidents, canonical problems, broken pages, titles and metas. Issues rank by severity, brand-new criticals raise an alert, and a manual run takes about 2 minutes.
☝ One red row in the crawler card - GPTBot blocked - and the criticals sorted above everything else. That's your week's technical to-do, already prioritized.
AI access
Four AI crawlers now matter to a local business: GPTBot (ChatGPT training), OAI-SearchBot (the one behind ChatGPT search), ClaudeBot and PerplexityBot. A blanket "block all bots" line a past developer added stays invisible until somebody checks - so every sweep checks.
Your robots.txt is evaluated the way Google evaluates it - longest match wins, Allow beats Disallow on ties - for Googlebot, Bingbot and all four AI bots. What we report is what the crawlers actually do, not what the file looks like.
Illustrative example - sample business.
The score
The health score isn't a vibe. Start at 100: each critical costs 20 points (capped at 60), each warning 4 (capped at 30), info items up to 5 total. Green at 85+, amber from 65, red below - with the delta since your last sweep right next to the gauge, so "better or worse" takes one glance.
Under the hood it's plain, careful HTTP - four concurrent fetchers, a 12-second timeout per page, redirects followed hop by hop. No paid APIs, no AI calls, no credits.
Illustrative example - sample business.
What gets checked
All four versions of your domain get probed - https and http, www and bare. Two live, non-redirecting versions raise a duplicate-site warning, and redirect chains of 3 or more hops get flagged, every hop counted (followed by hand, up to 6).
Weekly sweepGoogle-style longest-match parsing for Googlebot, Bingbot, GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot - each AI bot gets its own allowed or blocked verdict, plus sitemap and llms.txt checks.
Weekly sweepNoindex accidents on money pages, canonicals pointing at the wrong host, broken pages, and title and meta problems - the quiet mistakes that unrank a page without showing an error anywhere.
Weekly sweepLocation pages that are the same template with the city swapped get caught: pages 70%+ similar after every city name is stripped out are flagged as near-duplicates, with an on-demand AI rewrite to differentiate the weaker page (you confirm before it runs).
Weekly sweep · rewrite on demandOrphaned money pages - zero internal links but earning impressions - become warnings; pages with one link and real traffic become notes. The plan prescribes up to 10 target pages with up to 3 source pages each, and the suggested anchor text is that page's best real search query.
Refreshes with the weekly sweepEvery fetched page is read together, not one at a time: one business type in the schema or three, city pages whose structured data never names their own city, phone numbers and addresses that differ between pages, breadcrumb trails that disagree, and templates that have drifted apart (a skip link on 3 of 40 pages, no main landmark on 3). The copy-paste mistakes a per-page checker cannot see.
Weekly sweepChecks the five security headers scanners look for, and catches scripts and stylesheets that 404 together with the pages that still reference them. The sweep only ever requests pages and files the site links to itself, so it never looks like an attack to the host's firewall.
Weekly sweepAccreditation badges rendered as images that link nowhere, the same three photos on every city page, analytics installed with no phone or form events, response-time promises that differ between pages, images without dimensions, and CSS cached for a week with no version in the filename.
Weekly sweepCopy-paste LocalBusiness JSON-LD built from your verified profile data, FAQ markup drafted from the question-shaped searches your site already gets, and an llms.txt listing your top services and service areas by real clicks - no AI needed for that last one.
On demandThe numbers
Site Health, the internal link plan and the schema generators are all in the $69 flat plan. No site-audit add-on, no credits.
Start free - 5 days on us or see pricing →