Semrush vs Similarweb: keywords against traffic estimates

• What does this rival rank for, and which of their pages earns it? That is Semrush, and the answer is a list of queries and pages you can act on.
• How big is this company's audience, where does it come from, and how engaged is it? That is Similarweb, and the answer is a shape of visits across channels and countries.
• What do my own pages earn? Neither. That is Search Console and your analytics, free and exact.

Neither company can see another site's analytics. Both model, from different raw material, which is why the outputs differ in kind as well as in value.

• Semrush builds from its own view of search: a large set of tracked queries, the results pages for them, and a model of what a position on a query is worth in clicks. Its traffic figure is search traffic, estimated, not total visits.
• Similarweb models total traffic from panels, data partners and public sources, then extrapolates. It reports visits, the split by channel, geography, audience and engagement.
• Both get more accurate the larger the site. For a small product with a few thousand visits a month, both are rough, and Similarweb explicitly becomes rough at the small end.
• A limited free version of Similarweb exists, which is enough to see the shape of a large site.

Put one domain into both and you will get two numbers that do not reconcile. That is expected, and knowing why stops you trusting the wrong one.

• They measure different things: one search only, the other every channel. A site with a newsletter and a big social following will look far bigger in Similarweb.
• They model from different raw material: a query index against panels and partners.
• Coverage differs by market and by language. Both are strongest where their data is thickest.
• Neither sees brand-new pages quickly, and both lag reality by weeks.
• Neither can see the site's own analytics, which is the only exact figure and belongs to whoever owns the site.

• Your own impressions, clicks, positions and indexing state. Search Console has them, free, and they are Google's own figures.
• What visitors did after they landed. That is your product analytics, joined to the landing page.
• Revenue. No external tool has ever seen a payment.
• Why a rival ranks. The tools show the what; the reading of the page and the links is still yours.

Most small products need neither in the first months, and then need one of them for a specific week.

• Before writing a content plan: a keyword suite, for one or two months, to size the searches and see who holds them.
• Before a fundraise or a market sizing: a traffic estimator, to describe the shape of the market in numbers an investor recognises.
• Before a competitive launch: both for a week, then neither.
• For the routine month: Search Console, your analytics, and the hour you would have spent in a dashboard.

1. 1

   Look up a site you own

   Put your own domain into the tool and compare its estimate with your analytics. You now know which direction that tool errs in, for a site your size.

2. 2

   Look up a site you know well

   A partner, a former employer, anywhere you have seen the real numbers. Two data points are enough for a rule of thumb.

3. 3

   Compare rivals, never absolutes

   Rival A against rival B inside one tool on one day is a fair comparison. Either number quoted alone is not.

4. 4

   Check the market coverage

   If your buyers are outside the tool's strongest markets, halve your confidence again.

5. 5

   Write down the date

   Estimates are revised, models change, and a screenshot without a date is worthless in three months.