Competitor analysis
Five answers to one question: who is in front of you, and why
A competitor you audit is crawled and scored exactly like your own site, so the two numbers mean the same thing.
- panels, each answering a different question
- 5panels, each answering a different question
- competitors per site
- 3competitors per site
- crawler, page limit and check catalogue on both sides
- Samecrawler, page limit and check catalogue on both sides
- extra vendor calls - built from data already collected
- 0extra vendor calls - built from data already collected
Who owns the searches you track
Across the keywords you track, which domains keep appearing in the top ten and how high they sit - built from rank checks you already run, at no extra cost.
It surfaces competitors you never named. Often the site taking your traffic is not the one you were watching.
Which checks you actually differ on
Findings per hundred pages, check by check, so a small scan and a large one can be read against each other.
Split into where you are behind and where you are ahead - a list sorted only by size reads as either a scolding or a boast, and neither is useful.
And three more
Who the AI answers cite
The domains four assistants and Google's AI Overview name for your subject, counted by how many surfaces mention each - not by how many links one talkative vendor returned.
Topics they cover and you do not
Their pages grouped into subjects and matched against yours, with the closeness printed so you can judge it rather than trust it.
Health over time
One line per site, one point per day. A catalogue change is marked on the point where it happened instead of being hidden.
What they rank for, and what you do not
The panels above compare two crawls. This one asks the question people open a competitor page for and a crawl cannot answer: what are they winning that I am not? Type a domain and three answers come back at once - the searches they rank for with their position and the page that holds it, the searches they hold and you do not, and the domains the index says are actually in their results.
That third answer is usually the surprising one. Asked about one audit tool, the list came back with YouTube, Reddit, Google and Quora above the three rival tools anybody would have named - which is what the results pages actually look like, and not what a competitor list typed by hand ever contains.
It costs three keyword lookups from your monthly allowance, and the panel says so before you press it rather than after.
- They rank for
- 15,149searches, with the position and the page
- You do not
- 15,149the gap, sorted by how many search it
- Their real rivals
- 10domains the index puts beside them
The first two match here because the site asking ranked for almost nothing yet - that is what a gap looks like at the start. Three lookups, charged once.
Sample size is stated, not glossed over
A rate measured over eight pages is not a rate measured over five hundred, and the panel says which it is.
Where a conclusion rests on a zero from a small scan it is marked and pushed below the findings that were actually measured - a finding is evidence, and the absence of one on eight pages is not.
The same measurement on both sides
A competitor is crawled by the same crawler, under the same page limit, and scored by the same catalogue.
When two scans came from different catalogue versions the comparison says so and refuses to subtract them. A score from an older catalogue is not a worse site - it is a different measurement.
Questions
- Who holds the top of the searches you track; who the AI answers cite; the topics they publish that you do not; how both health scores have moved; and the individual checks where the two sides genuinely differ.
- Yes, and it is stated on the button: each audited competitor is crawled on your site's schedule and spends credits like any other scan. Three per site is the limit, for that reason.
- Yes. Each site keeps its own competitors and its own comparison.
- No, and the product says so on the panel. ahrefs.com passes one percent of its pages on structured data and carries 133 warnings per hundred pages, and it outranks nearly everyone. Technical health is table stakes, not a ranking model.
Start free
One site and 100 credits - up to 200 pages a scan, about 5 scans, and the whole report. No card. Paid plans add more sites, bigger crawls and the AI rewrites.