What a store visit conversion measures
This is a modelled number, and the word matters. Nobody scans a customer at the door and matches them to an ad click. Google takes the small share of signed-in users who have chosen to share location history, sees that some of them visited the mapped location after seeing or clicking an ad, and scales that sample up to estimate how many visits the campaign produced in total. What appears in the column is the output of that estimate.
Because it rests on a sample, it only appears when there is enough of one. Accounts need verified locations linked to the ads, enough ad interactions and enough measurable visits before Google will report it at all, and it is offered only in certain countries and for certain business types. If the column is empty, that is usually the reason rather than a tracking fault.
Why store visit conversions matter
For any business whose sale happens in person, online conversions tell only part of the story. A restaurant, showroom, clinic or bank branch can look unprofitable on form fills alone while the ads are quietly filling the room. A modelled visit figure gives that side of the account some weight in the reporting, so budget decisions are not made purely on what the website happened to capture.
It is also directional information about geography and timing. Which areas, hours and devices produce visits is often quite different from which produce clicks, and that pattern is more useful than the headline total.
Where store visit conversions go wrong
The commonest error is reading an estimate as a count. These figures should not be reconciled against a till roll or a door counter, and small day-to-day movements in them are noise. They also cannot be traced to an individual customer, so they will never explain who came in.
The second error is double counting. If a modelled visit is included in the main conversions column alongside form fills and calls, cost per conversion drops for reasons that have nothing to do with performance, and any bidding strategy chasing that column starts optimising towards a modelled number. Businesses with several branches close together also see the model struggle, because it cannot always tell which door someone walked through.
How to act on it
Keep it in a secondary column, watch the trend rather than the daily figure, and use it to compare areas and campaigns against each other rather than to prove a return in absolute terms. Make sure the location data feeding it is right: verified profiles, correct addresses and hours, and the location assets actually attached to the campaigns. When a real answer about offline impact is needed, a geographic holdout test is more honest than any model, and pairing the modelled visits with hard signals such as call tracking and voucher redemptions gives you something to check it against. Everything else in the account should still rest on properly configured conversion tracking.