Because they are measuring two different things, and neither is lying. A GPS tracker records a series of positions and adds up the straight lines between them; your odometer counts wheel rotations and multiplies by a tyre circumference the manufacturer decided on. A small, consistent gap between them is normal and expected — typically with the tracker slightly under the dashboard. A large gap is a signal worth investigating, and this article is about telling the two apart.
How does a GPS tracker actually measure distance?
It takes position fixes at intervals and sums the straight-line distance between consecutive fixes. That is the whole mechanism, and its properties follow directly from it. A road that curves between two fixes is recorded as the chord across the curve, never the arc, so every bend is measured slightly short. The more often the tracker samples, the closer the chain of straight lines gets to the real road — and the more battery it uses, which is the trade-off every tracking app is quietly making on your behalf.
Fixes also carry error. A phone reports a position with an accuracy radius, and in a city that radius can be tens of metres. A well-built tracker throws away fixes that are too uncertain to trust rather than summing the noise, because raw points in an urban centre drift through buildings and jump across junctions. Summing noise inflates distance; discarding it deflates distance slightly. Both choices are defensible, but only one of them is safe on a tax return, and the next-but-one section explains why.
How does an odometer measure it, and is it exact?
It counts wheel rotations and multiplies by an assumed rolling circumference, which makes it consistent rather than exact. The assumed circumference comes from the tyre size the vehicle was designed around. Fit tyres a size up or down, run them worn, run them soft, and every rotation covers a slightly different distance while the odometer keeps using the same multiplier. Worn tyres have a smaller rolling radius, so they over-count; a larger-diameter tyre under-counts.
There is also a deliberate bias in some vehicles, because speedometers are regulated so as never to under-read the true speed, and the same measurement chain often feeds the distance display. The practical upshot is that your dashboard is a perfectly good working record of how far the car went, and a poor choice of arbiter for a dispute measured in single figures. It is a reference, not a referee.
Where do the gaps actually come from?
From the places a phone cannot see satellites, and from the moments the phone was not asked to look. Both produce the same result — a stretch of driving with no fixes in it — and a tracker that bridges the gap by drawing a straight line from the last known position to the next one records the shortest possible version of whatever happened in between.
Some of these are unavoidable physics and some are fixable settings, and it is worth knowing which is which. A multi-storey car park will always be a dead zone. An aggressive battery optimiser killing the tracker every afternoon is not physics; it is a toggle, and it is the single most common cause of a tracker that is missing whole shifts rather than a few hundred metres.
- Tunnels and underpasses: no sky, no fixes, and the exit can be a long way from the entrance
- Multi-storey and underground car parks: common on supermarket and warehouse pickups
- Urban canyons: tall buildings bounce signals, so fixes arrive late, wide or not at all
- Cold starts: the first fix after the phone wakes can take a few seconds of driving to arrive
- The phone itself: battery optimisation, a killed app, location permission downgraded, or a dead battery mid-shift
Which direction of error is safe for a tax record?
Under-counting, without question — a log that is slightly short is conservative, and a log that is slightly long is a systematic overstatement of a deduction. That asymmetry is the whole reason a serious tracker prefers to discard doubtful data rather than include it. If you are ever asked to explain your mileage, "the recorded distance is the filtered driven distance, and it is if anything lower than what the car actually covered" is a comfortable sentence. There is no comfortable version of the opposite.
It is also why you should be suspicious of a tracker that reads high. Summing raw GPS points, counting the walk from the parking space to the customer’s door, or leaving a trip open while you sit in a restaurant car park all push the number up in a way that is invisible per trip and substantial across a year. The direction of a tracker’s error tells you a great deal about whether it was built by someone who has thought about audits.
What should I do when the two disagree materially?
Work out whether you are looking at measurement noise or missing trips, because the fixes are completely different. Noise shows up as a small shortfall spread evenly across every journey. Missing trips show up as whole absences — a Thursday with four recorded drives when you know you did fourteen. Open the log and look at a day you remember well; that single comparison usually settles it.
If it is missing trips, it is almost always the phone, and the checks below are in order of how often they are the culprit. If it is noise, leave it alone: a log that is consistently a touch under your dashboard is behaving exactly as it should, and adding a correction factor of your own invention would replace measured data with an estimate — which is the thing the whole log exists to avoid. Where a genuine trip was lost and you remember it, add it manually with a note, which is what manual entry is for.
- Battery optimisation: exempt the tracking app so the system stops killing it in the background
- Location permission: background tracking needs the always-allow setting, not while-using
- Precise location: an approximate-location toggle degrades every fix the app receives
- Power saving modes and OEM app-sleep lists, which are separate settings on many Android phones
- Storage and connectivity: a tracker that records offline should still be syncing once you are back on data
So which number goes on the return?
The per-trip log, because that is the number with evidence attached — and because neither the IRS nor HMRC has ever asked anyone for a photograph of their dashboard. A tax authority wants trips: date, distance, endpoints and business purpose, each one recorded at the time. An odometer reading is a single total that cannot be split into business and personal, which is precisely the split that matters. Use the dashboard as a sanity check on the log, not as a substitute for it.
BrewGig’s tracker is built to the conservative side of this trade deliberately: doubtful fixes are discarded rather than summed, trips end where the vehicle parked rather than where the walking stopped, and the route is stored beside the distance as the evidence for it. BrewGig is an independent product and is not affiliated with, endorsed by or partnered with any delivery or rideshare platform; all platform names mentioned here are the trademarks of their respective owners.