The argument is always about time
Strip a billing dispute down to its core and it's almost always about time. "Your guy showed up at ten, not nine like you promised." "He was only here twenty minutes — why am I paying for an hour?" "You billed me ninety minutes of drive time; there's no way it took that long." The tech remembers it one way, the customer another, the office has nothing but two conflicting memories and a quietly shrinking invoice. Field service runs on time — when the appointment started, how long the work took, how far the truck drove to get there — and yet most operations capture none of it precisely. They reconstruct it after the fact from a tech's recollection and lose the argument every time a customer pushes back.
A geofence fixes this by making arrival a recorded event instead of a memory. Draw a radius around the job site, and the moment a vehicle crosses into it the system knows the tech is there — and can stamp the time. Pair that with the photo evidence and on-site signature you're already capturing, and a visit that used to be a story becomes a defensible record: arrived at this time, was on site this long, drove this far to get here. The disputes don't just get easier to win — most of them stop happening, because there's nothing left to argue about.
What a geofence actually buys you
The point of a geofence isn't a cute "you've arrived!" notification. It's the operational and financial leverage that an automatic, accurate arrival timestamp creates:
- Honest on-site duration. Arrival time plus completion time is how long the tech was actually on the job — which is what the customer is really paying for. Captured precisely, it feeds straight into tracking the labor hours you bill instead of a round number someone guessed at the truck.
- Defensible drive time. A geofence on the depot and one on the site bracket the drive, so windshield time becomes a measured leg, not an estimate. That's the difference between billing travel you can prove and billing travel a customer can dispute away.
- A no-show and lateness record. When a customer claims the tech never came or came late, the geofence crossing — or its absence — settles it. It also quietly tells you which techs run behind, which is a scheduling signal, not a discipline one.
- One less interruption. The dispatcher who can see a tech has arrived doesn't have to call and ask, the same way GPS fleet tracking kills the "where are you" call. The geofence just makes the arrival itself the event.
None of this requires a tech to do anything extra. The crossing happens because the truck moved; the value falls out of work that was going to happen anyway.
The iron rule: suggest, never auto-apply
Here is the discipline that separates a geofence that helps from one that quietly corrupts your records. A geofence should suggest a status change. It must never silently make one.
The reason is that GPS is noisy. A signal bounces off a metal building and puts the truck a block from where it is. The tech parks at the far edge of a sprawling commercial lot, technically outside the radius while standing in the customer's lobby. A fix drifts during a long stop and registers a "departure" that never happened. If your system auto-applies an en-route-to-on-site transition off a noisy fix, every one of those glitches writes a wrong timestamp into the job record — the same record your invoice, your labor cost, and your dispute defense all depend on. You'd be corrupting your most load-bearing data with your least reliable sensor.
In Hosting Field the geofence is built around this rule on purpose. The site carries a latitude and longitude, and when a tech's GPS fix lands inside the radius the platform suggests the en_route → on_site transition — it surfaces a prompt, the tech confirms, and only then does the seven-state job FSM advance. The check never auto-applies the transition; the human stays in the loop and the FSM stays the single source of truth for what happened and when. The map informs the decision. It does not make it behind your back. That restraint is what lets you trust the timestamp the geofence produces — because a human confirmed the truck was really there, the machine just noticed first.
Setting it up so it actually works
A geofence that's set up carelessly trains everyone to ignore it, and an ignored prompt is worse than no prompt. A few things make the difference:
- Size the radius for the site, not a default. A tight 50-meter circle around a downtown address is fine; the same radius around a rural property with a long driveway means the tech is "on site" while still on the road, or never triggers at all. Set the radius to the site, and revisit it for the places that misfire.
- Geocode the site once, well. The geofence is only as good as the coordinates behind it. Hosting Field reverse-geocodes leg endpoints to real addresses and caches them, but a site whose pin is dropped on the wrong building will fire the wrong prompt forever. Fix the pin when a tech tells you it's off.
- Make confirming trivial, and dismissing honest. The tech should be able to confirm "yes, I'm here" in one tap — and just as importantly, to not confirm when the prompt is wrong. A suggestion that's easy to decline stays trustworthy; one that's annoying to dismiss gets reflexively tapped through, and then it's just noise wearing a timestamp.
The goal is a prompt the tech believes, because it's usually right and easy to override when it isn't. That trust is the whole asset.
What to measure
- On-site duration per job type. Now that arrival and completion are both real timestamps, the time the tech actually spent becomes a number — and a pattern. It's the raw input for catching the job types your estimates keep under-pricing.
- Geofence confirm rate. How often a suggested arrival gets confirmed versus dismissed. A high dismiss rate means your radii or your pins are wrong, and the prompts are training techs to ignore them — fix the data, not the people.
- Drive-time dispute rate. Track how often travel charges get challenged before and after you bracket drives with geofences. If the disputes don't fall, your customers aren't seeing the record — surface the arrival and departure times on the invoice or the customer status page.
A field visit is a sequence of times, and for most operations those times are stories everyone remembers differently. A geofence turns the most contested one — when the tech actually arrived — into a recorded fact, captured as a byproduct of the truck simply showing up. Hold the line that it only ever suggests the transition, set the radii to the real sites, and you trade a category of arguments you used to lose for a record you can stand behind. The customer stops disputing the time because there's nothing left to dispute.