How to check a competitor's web traffic, and how to read the estimate properly
Where competitor traffic numbers come from, which sources to use for which question, and the triangulation step that turns three estimates into a figure you can plan with.
Published by AI Knows Us · Updated 12 October 2026
Competitor web traffic is estimated, not read. Their analytics is private, so every tool builds a number from one of three inputs: a panel of real browsing behaviour, clickstream data bought from internet service providers and browser extensions, or a model that multiplies the keywords a site ranks for by an expected click through rate. The way to get a figure you can plan with is to take estimates from two or three sources that use different inputs, compare them, and use the result for relative scale rather than as an exact count. This page shows which source answers which question and how to run the check in order.
This page covers estimating traffic for a site you do not own. For your own numbers, see how to measure organic traffic properly.
The three ways a traffic estimate is built
- Panel and clickstream. A large sample of real sessions is scaled up to the whole web. This covers every channel, so it is the only approach that shows direct, social and referral traffic as well as search. Accuracy is best on large sites with plenty of sample.
- Search visibility models. The tool knows which keywords a domain ranks for and at which positions, applies a click through curve, and sums the result. This is precise about search and silent about everything else. It is strong on small and mid sized sites in competitive niches.
- Publisher and platform data. Some figures come from the site itself: an advertising media pack, an investor deck, a press release, an app store listing. When it exists it is the best input you will get.
Knowing which input a number came from tells you immediately what it is useful for. A panel figure answers how big is this business online. A search model answers how much of that comes from queries I could compete for.
Which source for which question
| Source | Input it uses | Question it answers well | Free view |
|---|---|---|---|
| Similarweb | Panel and clickstream | Total visits, channel split, country split, audience overlap | Limited free summary per domain |
| Semrush | Both, in separate reports | Organic traffic from keywords, plus a traffic analytics view for all channels | Capped free queries with an account |
| Ahrefs | Search visibility model | Organic traffic value, top pages, which keywords drive it | Free site explorer views for a verified site |
| The site itself | First party | Audited figures, where a media pack or filing exists | Free if published |
| Google Trends | Relative search interest | Direction of brand demand over time, not visit counts | Free |
For how the different ranking and traffic indexes build their league tables, see how web rank tools build their numbers, and for the two different figures inside one suite, the two Semrush traffic numbers and which to use.
The check, in order
- Write the question first. Are you sizing a market, choosing which competitor to study, finding content gaps, or building a board slide? The question decides which number you need.
- Get a panel estimate of total monthly visits and the channel split. This tells you whether the site lives on search, on paid, on social or on direct brand demand.
- Get a search model estimate of organic traffic and export the top fifty pages by traffic. That export is the most useful artefact of the whole exercise.
- Compare the two. If the panel total is far above the search estimate, their growth is not coming from search and copying their SEO will not reproduce it. If the two are close, search is the engine.
- Sanity check against something physical. Branded search volume, review counts, headcount on their careers page, store locations, app downloads. Online traffic that is wildly out of step with the size of the business is a signal to look again.
- Record the date and the tool. Always compare this month to last month in the same tool. Mixing tools across months produces movement that is really a different crawl.
- Turn it into a decision. The output is a shortlist of their pages and queries worth answering better, not a number in a spreadsheet.
What to do with the top pages export
This is where competitor traffic work pays for itself. Take their fifty highest traffic pages and sort them into four buckets:
- Pages you have no equivalent of. Direct content gaps. Judge each one by whether the query belongs to a buyer or a browser.
- Pages where you have something weaker. Usually the fastest return, because improving an existing page is quicker than launching one.
- Pages that pull traffic you do not want. Glossary and definition pages that never buy anything. Note them and move on.
- Pages that reveal the offer. Pricing, integrations, locations, comparison pages. These tell you what they think buyers ask before choosing.
Run the same exercise for three competitors and the overlap is your plan. The structured version of this is covered in a competitor analysis method that ends in decisions.
Making it a monthly habit instead of a one off
A single reading tells you where a competitor is. A series tells you where they are going, which is the more valuable thing. Keep it light enough to actually do:
- Three competitor domains, one tool, the same day each month.
- Four numbers per competitor: estimated visits, estimated organic traffic, number of ranking keywords, new pages published.
- One line of commentary on anything that moved more than you expected.
- Alerts for their new pages and their paid search ads, so a launch does not reach you through a sales call.
The tooling for the watching side is set out in monitoring competitors in search.
Can I see a competitor's exact website traffic?
Exact session counts live in their own analytics. What you can get is a well built estimate from two or three independent sources, which is enough to size the business, compare competitors against each other and find the pages worth answering.
Which free tool is best for checking competitor traffic?
Start with the free Similarweb summary for total visits and channel split, add a capped free view from Semrush or Ahrefs for the organic side, and read both together. The two inputs are different, so agreement between them is meaningful.
Why do two tools give completely different numbers?
Because they are measuring different things with different inputs. One scaled a behavioural panel, the other multiplied keyword rankings by a click curve. Use the gap between them as information: it tells you how much of their traffic is search.
How accurate are competitor traffic estimates?
They are at their best on large sites with plenty of sample and on search heavy sites in well covered niches. Treat them as reliable for relative comparison and direction over time, and use the triangulation step before you put a single figure in front of a board.
Is checking a competitor's traffic allowed?
Yes. Every method here uses publicly available or commercially licensed data and published search results. You are reading the outside of their website, which is what it is published for.
Triangulation with a fourth source
You triangulate three traffic estimates into one figure. A competitor's visits that began with an assistant are in none of them, because a person who is told a name and then types it in looks like direct traffic. Ask the assistants about your category and note who gets named; it is a rough guide to who is gaining that unseen traffic. Add it as a line in the monthly habit. AI Knows Us records the names and the sources.