Read competitor traffic estimates honestly

No tool can see a competitor's analytics. What a traffic estimate does instead is list the terms a domain ranks for, attach a search volume to each one, multiply through by a click rate assumed for that position, and add the rows up. Every error in those three inputs passes straight into the headline number.

That does not make the estimate useless. It makes it an instrument with known distortion, and you can correct for the distortion once you know where it sits. Treated as a measurement, it will send a quarter of work after a gap that was never there.

Where the headline number comes from

The figure a domain overview leads with is assembled, not observed. Rank data comes from a crawler that checked the result page. Volumes come from a search-demand index. Click rates come from a curve fitted to positions. Multiply, sum, publish.

Two consequences follow, and both matter more than the number itself.

The first is that two tools will disagree about the same domain, because each one holds a different index, sources volume differently, and assumes a different curve. Wide disagreement does not mean one tool is broken. It means the true figure is uncertain, and that is the honest thing to report to whoever is asking for a number.

The second is that a search-only estimate cannot see anything but search. A competitor taking a large share of their demand from short video, a marketplace, a newsletter, or an app storefront will look smaller than they are. The estimate can also run the other way, overstating a domain whose rankings sit on terms nobody clicks. Either direction is enough to wreck a benchmark built on the headline.

One domain, several businesses

The single largest distortion is also the easiest to correct. A domain-level total rolls every business line that domain operates into one figure.

Look at the terms carrying the most estimated traffic for an incumbent in your category and the pattern is common enough to expect: alongside the terms you compete on sit the terms for adjacent services, for a retail arm, for a brand of consumables, or for a content site the company runs. Those are real visits arriving at real pages. They are not visits a competitor could take from you, and they are not visits you are losing.

Before benchmarking against any total, split it. Filter the ranked-keyword list to the terms describing what you actually sell, and total only those. For most markets what survives is far smaller than the headline, and the shortfall you had been planning against shrinks along with it.

Why keyword counts run high

The second distortion inflates the keyword count rather than the traffic figure, and it comes from how search volume is published.

Volume data is grouped. Search platforms report a cluster of near-identical queries — plurals, reordered words, small rewordings — as one number, and attach that number to every member of the cluster. A downstream model that reads each keyword as an independent row and sums their traffic estimates will count the same pool of demand several times over.

Watch for the signature: a run of rows reporting identical volume, identical estimated traffic, and the same ranking URL. Those are not separate opportunities. They are one opportunity described five ways.

This is why a reported keyword count is a weak measure of how far ahead a competitor is. A domain showing tens of thousands of ranked terms may hold a few hundred genuine pools of demand, grouped and re-reported.

Read pages, not keywords

Both distortions clear up when you switch from the keyword view to the page view. Grouped variants resolve to a single URL. Other business lines become obvious. And pages map to work: a keyword is not something you can produce, a page is.

Three questions do most of the sorting:

1. How concentrated is the traffic? If the top handful of pages carry most of the estimate, the advantage is shallow enough to see all of it at once. If the estimate is spread across hundreds of pages, the domain has depth.
2. Which of those pages are in your market? This is where the multi-business correction happens.
3. How good and how current are they? A thin page holding a strong position for a query that matters to you is an opening, and the cheapest one on this list.

An estimate answers none of these. The keyword and page rows behind the estimate answer all three, which is the real reason to pull the data rather than screenshot the headline.

Do it with an agent

Connect the Doverity MCP and the whole correction runs in one pass, with the agent showing its working rather than handing over a number.

Using the Doverity MCP, give me a defensible read on the search
footprint of [competitor.com].

1. Call get_domain_overview. Report estimated organic traffic and the
   organic keyword count, and label both as modeled.
2. Call get_ranked_keywords for the same domain, sorted by traffic.
   Find groups of rows that report the same search volume, the same
   traffic estimate, and the same ranking URL. Estimate the share of
   the headline sitting in those groups.
3. Group all keywords by the URL that ranks for them, then rank the
   resulting pages by their traffic contribution, twenty rows deep.
4. Sort those pages into [what I actually sell] and every other
   business line, and total the traffic on each side.

Finish with two figures: the headline estimate, and the share of it that
is actually aimed at my market. Then one paragraph on what you are least
sure about and what would change your answer.

That closing paragraph is not decoration. Hand over an estimate with no statement of its own uncertainty and whoever receives it will read the figure as fact. That is the failure this page is here to prevent.

Competitor traffic estimate FAQ

Can you trust a competitor traffic estimate?

Reliable as a direction, unreliable as a figure. They are assembled from ranked keywords, modeled volumes, and assumed click rates, so they carry the error of all three, and they fold every business line on a domain into one total. Read them as a magnitude and a trend. Never plan revenue against one.

Why does each tool show a different traffic number?

They hold different indexes, draw volume from different places, collapse near-identical queries on different rules, and fit different click curves. Every one of those choices moves the total. Agreement between two tools is reassuring; disagreement is normal and should be reported as uncertainty rather than as one tool being wrong.

Why do several keywords show the same search volume?

Because volume is published per group of near-identical queries rather than per query. Every member of the group carries the group's number, so a model that sums per-keyword estimates counts one pool of demand more than once. Grouped rows with identical volume and an identical ranking URL are the tell.

Is organic traffic the same as total traffic?

No. An organic estimate covers search results only. Direct visits, email, paid, social, and referral are invisible to it, and for plenty of businesses those channels carry most of the demand — which is why a modest estimate and a strong business are not in conflict.