Estimating a site’s traffic without access to its analytics is more an exercise in deduction than in guesswork. Before opening any dashboard, it helps to be clear about what can and cannot be known. This checklist gathers the steps that do provide useful signals, ordered from most reliable to most speculative.
Before you start: what to expect and what not to
Without access to the server or an analytics account, you won’t get an exact figure for visits. What you can get is an order of magnitude and a relative comparison between two domains. Any number that comes out of these methods is an estimate with a wide margin of error, and it should be treated as such. If you need data for an important business decision, this approach is not enough.
Verification checklist without paid tools
1. Check the source code for trackers
Open the page with Ctrl+U or with the browser inspector and look for analytics identifiers: gtag, analytics.js, googletagmanager.com, hotjar, plausible, matomo. If the site has no tracker installed, the owner doesn’t know its real traffic either, and third-party estimates will be even less reliable. If it has several installed, it’s a sign of a team that measures with a certain maturity.
2. Check whether the domain hides a testing subdomain
A common mistake is measuring the staging environment believing it to be production. Try variants such as dev., staging. or test. in front of the domain. If they respond and aren’t blocked by authentication, that residual traffic can distort any external estimate based on crawling.
3. Check the sitemap file and robots.txt
The number of URLs included in the sitemap gives an idea of the size of the catalog. A site with 50 pages and another with 50,000 don’t compete in the same league. It isn’t traffic, but it narrows down the potential ceiling. Note: many sitemaps include category and filter pages that don’t generate visits.
4. Use the cache and Wayback Machine history
Comparing the current homepage with the one from two years ago reveals whether the project has grown, changed focus or been abandoned. A blog that used to publish three times a week and hasn’t been updated in eight months has usually lost organic traffic, even if you can’t quantify it.
5. Look for public mentions of figures
Press releases, interviews, “about us” pages and investor profiles sometimes publish audience data. They are self-interested statements and hard to verify, but they serve as a starting point. Always cross-check with a second source.
6. Analyze activity on social media and newsletters
The number of visible interactions on social profiles doesn’t translate directly into web visits, but a channel with thousands of followers and posts with dozens of reactions indicates little real reach. It’s a weak signal: use it only as context.
7. Check presence in search results for brand keywords
If searching for the brand name brings up many indexed pages with varied titles, the site has enough content to capture navigational searches. If only the homepage appears, the footprint is minimal.
What this method doesn’t detect
None of these steps reveals internal traffic (logged-in users), traffic from mobile apps, or traffic from paid campaigns active at the time of the query. Nor does it distinguish between human visits and bots, and many automated crawlers generate requests that inflate any indirect count. According to industry documentation, a significant portion of the traffic recorded by third-party tools corresponds to non-human activity.
One specific case that comes up repeatedly: analyzing an ecommerce site with thousands of filter and parameter URLs. The sitemap shows a huge volume, but most of those pages receive no visits and are intended only for crawling. Estimating traffic from the size of the sitemap in that scenario leads to overvaluing the site by a factor of ten or more.
Practical conclusion
This checklist is useful for ruling things out, providing context and comparing qualitatively. If you need a figure to negotiate a collaboration or evaluate a purchase, ask for access to the owner’s analytics or use a professional tool with panel data. Without that, any number is a hypothesis, not a fact.