Verified vs estimated data

Not every number in ASO Atlas has the same level of certainty behind it, and pretending otherwise would be a disservice. This article explains the two confidence levels behind the popularity score, why estimates exist at all, and how to make decisions with each.

Most popularity scores are verified

The large majority of popularity scores you see are backed by verified Apple search data: a real measure of how often people search a phrase on the App Store, on the same 0 to 100 scale you see everywhere in ASO Atlas. For head terms and established keywords this coverage is essentially complete, and the numbers are accurate enough to build a metadata strategy on.

Where estimates come in

No tool on earth has verified demand data for every phrase a person might type. Very new terms, narrow long-tail phrases and niche local queries sometimes fall outside verified coverage. When that happens, ASO Atlas has two honest options: hide the keyword, or tell you what we believe its demand is. We do the second.

For those keywords we show our calibrated estimate: a score produced by our own model and continuously tuned against the verified data, so that an estimated 40 means roughly what a verified 40 means. It is an approximation and we treat it as one. What we never do is invent a precise-looking number and present it as fact, or silently drop a keyword just because the certain answer is not available yet.

Estimates upgrade automatically

An estimate is usually a temporary state, not a permanent one. When verified Apple search data becomes available for a keyword, its score upgrades automatically in the background, and no action is needed from you. In practice this often happens shortly after you first research a term, so a phrase that starts out estimated can be showing a verified number by the next time you look.

This is also why a score can shift slightly a day or two after you first see it. That movement is the score getting more certain, not less.

Difficulty has no verified or estimated split

Keyword difficulty works differently. It is computed fresh from the live storefront leaderboard: the actual apps ranking for the phrase right now, how many of them target it in their name, and how established they are. Because the input is the current state of the search results themselves, there is nothing to estimate. Every difficulty score has the same footing, whether the popularity next to it is verified or estimated.

How to work with each

A practical rule of thumb:

• Treat estimates as directionally right. They are good enough to shortlist with: separating "nobody searches this" from "real demand exists" is exactly what they are calibrated for.
• Let big decisions rest on verified numbers. Rewriting your title around a single keyword, or dropping a long-standing one, deserves the certain figure rather than the approximation.
• Re-check shortlisted keywords after a day or two. Since estimates upgrade on their own, a quick second look often turns your shortlist's approximations into verified numbers before you commit any metadata characters to them.

Estimates keep your research complete; verified data makes it dependable. Used in that order, you get both.