How GPS Door-Knock Verification Works: Two Radii, Two Jobs

Every door-to-door sales app uses the phone's position somewhere. The obvious design is one radius: if the rep is within some distance of the lead's address, the visit is real. In Dromo, the field sales app we build, that one number turned out to be doing two jobs, and the two jobs want very different answers.

Two questions that look like one

The first question is which house is this. When a rep walks up to a door, the app offers to name it: "this is Ruth Baker's house, 3rd time here". That prompt is a claim about one specific address, and the rep will usually take it at face value and tap an outcome.

The second question is does this knock count. That is a question about the rep's day and pay: was the rep genuinely out in the territory, working the book, when the visit was logged? It is a management rule, not a navigation aid.

Put both behind one radius and you have to pick which question to get wrong.

Naming a door: 60 yards

A residential lot is 50 to 80 feet wide. At 60 yards, the circle around a rep covers the house they are standing at and its neighbours, not the block. That is as far as the app will go to name a house.

It also has to trust the fix before it asks. The phone's own position must be accurate to 55 metres or better and no more than two minutes old. A worse fix names no house at all. That is deliberate: a wrong address logged is worse than no prompt, because it quietly attaches a visit to the wrong homeowner and nobody finds out until the follow-up goes to the wrong door.

When the fix is good and the house is close, the door card does the rest: it says how many times the house has been tried, fills the office's own opener in with the homeowner's first name, the rep's name and the street, and lays the outcomes out as buttons. One tap logs the visit and the next door within reach comes up.

Counting a knock: 250 yards, set by the office

Whether a knock counts is decided by the back office, per rep, 250 yards by default. It is wider on purpose. GPS drifts between houses, under trees and along long driveways, and a rule that decides pay should not punish a rep for the phone's bad afternoon.

It also belongs to a different owner. The rep's phone decides what to show on a porch. The office decides what counts, with explicit outcomes and reason codes behind every decision, so a disputed visit can be explained rather than argued.

Why split them

At 250 yards, the naming prompt would put the wrong address in front of a rep. At 60 yards, the counting rule would reject real work whenever the signal wandered. Each number is right for its own question and wrong for the other one.

The general version of this is worth keeping: when one threshold is doing two jobs, the jobs usually want different values, and the fix is not a better number. It is two numbers, each owned by the part of the system that is accountable for it.

Dromo is our field sales app and CRM; the rebuild behind this is written up in Dromo 2.0, and the product is at statistique.io/dromo.