Site
Your own site as the engines meet it — what was crawled, what the markup says, and who is allowed to read it.
Every other page in Recomma is downstream of an answer. This one is not: it is the state of your own site whether or not a single chat has been sampled, and whether or not anybody has decided to act on it.
It reads from two things the product was already collecting and neither of which had a home: the crawl of your sitemap, and the audit findings behind Site health tickets. Findings could previously only be reached through a ticket, which meant the state of your site was legible only where somebody had already decided to do something about it.
The strip
Three figures across the top, each the worst number in the section under it:
| Figure | What it reads |
|---|---|
| Pages read | Fetched against discovered, with the failures and the backlog beneath it |
| AI crawlers shut out | How many are blocked outright, and how many more are kept off some paths |
| Markup faults | Failing pages in total, over how many rules, and how many are ticketed |
Pages are found from your sitemap rather than from your citations, because the pages with problems are usually the ones nothing has quoted yet. It is a backlog worked on the schedule, so a waiting count above zero is normal on a large site and is the reason an audit can be a true statement about part of a site.
Crawlability
What your robots.txt says about the crawlers that fetch pages in order to
answer questions — GPTBot, ClaudeBot, PerplexityBot and the rest.
Only findings are listed. Most sites carry housekeeping disallows that every crawler on the web falls under, and reporting those would say every site blocks every agent.
A row is a decision, not a crawler. A robots.txt is a short list of
decisions — “keep these bots out of that path” — so one rule that shuts nine
crawlers out of one path is one row here rather than nine rows repeating the
same path. The crawlers it governs are listed under it as chips, grouped by
the company that runs them, which is how anybody thinks about who is allowed
in; hover a chip to see the tokens behind it.
- Blocked — shut out of the site.
- Partly blocked — shut out of some paths, which are listed on the row.
What changed
robots.txt is re-read on a schedule, and every change is kept. Open the fold
at the foot of the panel for the history: which crawler moved, which way, and
on what day. A brand that lost its visibility the week somebody added a
Disallow has the answer here rather than a coincidence.
Blocking a training crawler is a decision, not a fault
Only the crawlers that fetch a page in order to answer a question matter here. Keeping CCBot or Bytespider out costs you nothing in the engines this product measures — they are listed because a reader asking “who can read my site” wants the whole answer, not because they are a finding.
If robots.txt could not be fetched, the panel says so rather than reporting
the site as open. Unreachable is not permissive.
Markup
Every audit rule the site currently fails, with how many pages fail it. Open a row and it gives the rule in a sentence, the pages it fires on, and the markup on each one.
The State column is the link back to work:
- ticketed — an action is open against this rule on Opportunities.
- untriaged — no ticket. Which is not the same as no problem: a ticket is raised when a fault clears an impact bar, and a rule failing on four pages of a large site may never clear it. Somebody reading their own site should see it either way.
That distinction is why this page is read-only. A ticket is work somebody has decided to do; this is the site’s condition whether anybody has decided anything, and the two answer different questions. Run the audit, take tickets and mark them done on Opportunities.
Pages
The crawled inventory, worst first: pages that failed, then pages nothing has read yet, then the oldest readings. Each row carries the URL and the status the crawler got back.
The table pages through the whole inventory rather than showing a sample of it.