Every ranked-keyword dataset, at every price, carries the same four caveats. Knowing them is the difference between using this data well and being misled by it.
The positions are an index snapshot, not a live search. The tool looks the domain up in an aggregated index of keywords and the pages ranking for them, rather than running a search at the moment you press the button. A ranking that moved this morning may not have landed yet. That is close enough to decide what to work on and not close enough to report — anything going in front of a client should be confirmed with a live rank check first.
Search volumes are annual averages. The figure beside each keyword covers the previous twelve months, so a term that tripled and a term that halved can report the same number. Volume tells you size, never direction.
Personalisation and location mean there is no single ranking. The position shown is a depersonalised result for the country and language you selected. Two real users in different cities, signed in with different histories, may both see something else. Check the markets you sell in — all 213 of them are available — rather than assuming a US snapshot describes the world.
Rankings are not traffic. A domain with more keywords is not automatically a domain with more visitors; distribution matters more than count, and a single page holding position 1 on a high-volume term can outperform a thousand page-three rankings. If traffic is the question you're actually asking, estimate a domain's organic traffic rather than inferring it from list length.
One last point, and it's the one that has changed most recently. Competitor visibility is no longer only a question of who ranks. Across the category prompt sets we track, the top three brands captured 57.6% of all AI mentions in July 2026 — being fourth in a category is much closer to invisible than to the podium. Ranked keywords remain the entry ticket to that surface, per the 91.3% figure above, but the surface itself concentrates far harder than a results page does. When you study a competitor's keywords now, you are studying the input to both.