Unified audit · one snapshot
One crawl and one sample of Perplexity, Gemini and DeepSeek, read five ways: site health, search visibility, AI visibility, agent readiness and the fix list. Same snapshot, same timestamp, so the six answers cannot contradict each other.
Enter a domain, like example.com.
Free: 1 Perplexity sample × 1 tracked question · no signup · report in about 90 seconds
The report · five lenses
Every audit reads the same snapshot in the same order. Each block below is a chapter of the delivered report: the question it settles, and the three things you will see in it.
Whether the page can be fetched, indexed and served quickly, judged from the crawler side rather than through a browser.
Where the page ranks for the questions buyers already type, and which of those questions have no page behind them.
What Perplexity, Gemini and DeepSeek say about you, stored verbatim with the prompt and the sources they cited.
Whether a crawler that never runs JavaScript can read, parse and quote the page without help.
The shortest path from a finding to a change someone can ship, ordered by what moves citation rather than by checklist order.
Why merge the three
Run an SEO tool, an AI visibility tool and a technical crawler on the same site in the same week: three numbers, three timestamps, three explanations. The argument that follows is about the tooling, not about the site.
01
Every page and every prompt read at the same moment.
The crawl, the field data and the engine sample share a timestamp, so a finding on one lens can be checked against the other four without asking when each of them was taken.
Same snapshot · 1 timestamp, 5 lenses
02
A drop has exactly one place to look.
When AI visibility falls and site health holds, the cause is a source or a prompt, not a broken page. With three separate tools you first have to decide which scoreboard to believe.
Same sample · 3 engines, one prompt set
03
The only comparison left is your site, month over month.
Different tools crawl with different user agents, on their own schedules, against their own prompt lists. Re-running all three to reconcile them costs a day and still leaves three baselines.
Same baseline · month over month
Sampling and cost
No install, no DNS record, no access to your analytics. A run reads only what anyone can read from outside, and it stores the evidence next to every number.
Perplexity, Gemini and DeepSeek by default. Additional engines can be enabled per language.
A full run samples at least four prompts. The free audit samples one question per engine.
After 30 days the snapshot is re-taken, never re-labelled as current.
Small sites finish in about 90 seconds. The crawl budget is capped, so a large site truncates instead of hanging.
Cost is one line on the invoice: every paid plan watches 500 tracked questions across up to 100 domains, extra capacity is bought by credit, and credits do not expire. The prompt library holds 210 questions across six industries.
Questions
A snapshot stays valid for 30 days. Paid accounts are re-measured on their plan cadence, daily or weekly, and every run stores its own timestamp, so two reports are never silently mixed. A free audit samples once per domain and keeps that snapshot.
Site health, the agent readiness checks and one sampled question per engine, across the sampled engines. The rest of the prompt matrix stays locked, and no card is needed to read the free report.
Yes. Every audit has a clean public URL at /audit/yourdomain.com/ that can be sent to a client or a colleague without a login, and the public report is indexable. Private projects, prompt libraries and alerts stay behind the account.
The full audit report downloads as HTML, with a PDF when the renderer succeeds. For now a report lives at its public URL, and the fix list arrives as files you can commit to your own repository.
Free · no signup