When Your Mileage App and Your Odometer Don't Match: Reconciling GPS Data for the IRS (2026)
Published: September 3, 2026 ยท Reading time: 8 min
TL;DR: A GPS mileage app's yearly total and your vehicle's actual odometer delta answer two different questions โ total miles (odometer) and business miles (the log) โ and when the app's full-year total doesn't add up to the odometer delta, real driving happened that the app never recorded. That gap is either unlogged personal miles (no deduction lost) or unlogged business miles (a real deduction sitting unclaimed), and because vehicles are listed property under IRC ยง280F, the heightened substantiation rules of ยง274(d) block the Cohan rule's usual estimation fallback โ you can't just round the gap away. The fix is a five-step reconciliation: nail down the odometer-based total, pull the app's full summed total, isolate the gap, reconstruct what you can from calendar and receipt evidence, and document the reconciliation itself as part of your record.
Every freelancer who tracks mileage with an app eventually hits the same unsettling moment: the app says one number, the odometer says another, and the two don't reconcile. It's tempting to just trust whichever number is more convenient. Neither move is right โ the gap itself is information, and running it down usually strengthens your deduction rather than shrinking it.
Two Numbers, Two Different Jobs
It helps to separate what each figure is actually for:
- The odometer delta (year-end reading minus year-start reading) is your total miles driven for the year โ the figure Schedule C Part IV asks for directly, and the only objective source for it.
- Your GPS app's log is where your business miles come from โ the trips you've classified as business, summed for the year.
Business-use percentage = business miles รท total miles. For this to be credible, the app's full recorded total (business trips plus any personal trips it also logged) should roughly match the odometer delta. When it doesn't, the app didn't capture everything you actually drove โ and that missing driving is exactly what the reconciliation process below is for.
Why the Gap Happens
The app under-counts (most common)
- Phone battery died mid-day, or was left charging at home during a drive
- The operating system killed the app in the background โ both Android's battery-optimization settings and iOS's background-activity limits are known to silently stop location tracking on apps that haven't been opened recently
- Location permission was revoked after an OS update, without the driver noticing until months later
- The app hadn't finished launching before the drive started, missing the first few minutes or the whole short trip
- A multi-day stretch was simply never logged because the driver forgot to open the app
Every one of these is a genuine hole in the record โ miles were driven, and nothing captured them.
The app over-counts (less common, but real)
- GPS drift in dense urban canyons or near large buildings can register phantom movement from a stationary phone
- Motion-based auto-detection can misclassify travel as a passenger, on a train, or on a plane as a "drive"
- A duplicate-logged trip โ the app restarts mid-drive and records the same trip twice
- A misread odometer figure (a transposed digit on the year-start reading) can make an otherwise-accurate GPS total look inflated by comparison, when the real problem is the odometer figure
The direction of the mismatch tells you which category to investigate first.
Why You Can't Just Round the Gap Away
Some Schedule C expenses tolerate a reasonable estimate when records are incomplete, under the Cohan rule (Cohan v. Commissioner, 39 F.2d 540 (2d Cir. 1930)) โ a court can allow an estimated deduction even without perfect documentation, as long as there's some credible basis for the estimate.
Vehicle mileage doesn't get that leniency. Vehicles are listed property under IRC ยง280F, and IRC ยง274(d) imposes heightened substantiation requirements specifically for listed property โ requirements the Tax Court and the IRS have consistently held override the Cohan rule entirely for this category. In practice: you cannot simply estimate a plausible-sounding number to fill a mileage gap. You need either a reconstructed entry backed by real corroborating evidence, or you need to concede the miles are unaccounted for and leave them out of your business total.
The Reconciliation Workflow
Step 1: Lock in your total-miles number
Use your actual odometer readings from January 1 and December 31. If you didn't record them, reconstruct from repair invoices, oil-change stickers, or state inspection records that print the odometer reading with a date โ see Odometer Readings and Your Mileage Log for the reconstruction method in full.
Step 2: Pull the app's full-year summed total
Export every trip the app logged, business and personal alike, and sum the mileage. Comparing the odometer delta against only your business-classified trips will always show a "gap" โ that's expected, since personal driving isn't in the business total. The real comparison is odometer delta vs. everything the app recorded.
Step 3: Calculate and diagnose the gap
Worked example: A freelance photographer's odometer went from 31,400 to 45,600 over the year โ a 14,200-mile total. Her mileage app's full-year export (all trips, business and personal) sums to 9,400 miles.
14,200 โ 9,400 = 4,800 unlogged miles.
That's a 34% gap relative to her odometer total โ large enough that it isn't measurement noise. Something real happened to those 4,800 miles.
Step 4: Reconstruct what you can, exclude the rest
She checks her calendar and finds: a two-week vacation drive to visit family (personal, no reconstruction needed, accounts for roughly 1,600 miles based on the trip's known round-trip distance) and a six-week stretch in the spring when her phone's OS had silently disabled the app's background location permission after a system update โ a period that included eleven confirmed client site visits, each corroborated by a calendar entry and a client invoice showing the date and address.
Using mapping-service distance estimates between her home office and each client address, she reconstructs 1,340 business miles for that six-week gap, entering each as a dated log entry noting the reconstruction source. The remaining 1,860 miles of the original 4,800-mile gap she can't positively account for โ she leaves them out of both her business and personal totals rather than guessing, which is the conservative, defensible choice IRC ยง274(d) requires.
The app's original 9,400-mile total already split as 6,200 business and 3,200 personal. Here's how the full 14,200-mile odometer total now breaks down:
| Category | Miles |
|---|---|
| Business miles, originally logged by the app | 6,200 |
| Personal miles, originally logged by the app | 3,200 |
| Business miles, reconstructed from the six-week gap | 1,340 |
| Personal miles, identified from the gap (vacation) | 1,600 |
| Unaccounted for โ left out of both totals | 1,860 |
| Total (matches the odometer delta) | 14,200 |
Her corrected business-mileage total is 6,200 + 1,340 = 7,540 miles โ a real recovery of 1,340 miles she would otherwise have simply lost, worth $971.50 at the 2026 first-half standard mileage rate of $0.725/mile (1,340 ร $0.725 = $971.50).
Step 5: Document the reconciliation itself
Write a dated note summarizing what you found: the size of the gap, the cause you identified (the OS permission issue, in this example), and how each portion was resolved. This note isn't busywork โ in an audit, a documented reconciliation is itself evidence that the log was carefully maintained, which is one of the credibility factors an examiner weighs. A driver who can produce "here's the gap, here's why, here's how I fixed it" looks far more credible than one whose numbers simply happen to line up with no visible process behind them.
What This Isn't a License to Do
Reconciliation is not an invitation to pad a shortfall with convenient-sounding business trips you can't actually corroborate. Every reconstructed mile still needs a real, checkable source โ a calendar entry with a specific client and address, an invoice, a receipt with a location and date. "I probably drove to a few more job sites that month" is not a reconstruction; it's a guess, and IRC ยง274(d)'s listed-property rules exist specifically to block guesses like it.
Authoritative References
- IRS โ Publication 463: Travel, Gift, and Car Expenses
- Cornell LII โ 26 U.S. Code ยง274(d): Substantiation requirements
- Cornell LII โ 26 U.S. Code ยง280F: Limitation on depreciation for luxury automobiles; limitation where certain property used for personal purposes
- Cohan v. Commissioner, 39 F.2d 540 (2d Cir. 1930)
Related reading: Odometer Readings and Your Mileage Log ยท GPS Mileage Tracking Apps & IRS Compliance ยท 2026 IRS Mileage Rate
Tired of reconstructing gaps by hand at tax time? Start a free CentSense account and every receipt and trip gets logged and reconciled as it happens โ so a dead phone battery in March doesn't cost you a deduction in April of next year.
This guide is general education for U.S. freelancers and Schedule C filers in 2026. It is not personalized tax advice โ bring your specific situation to a CPA or EA.
Related reads
Continue learning with more tax and expense guides for freelancers.
2026-09-05
Schedule C and Self-Employment Tax for Bona Fide Residents of Puerto Rico (2026)
2026-09-05
Locum Tenens Physician Tax Deductions (2026): The Complete Schedule C Guide
2026-09-05
Disabled Access Credit (Form 8826) vs. Barrier Removal Deduction (ยง190): Stacking Both for an Accessible Studio (2026)
2026-09-05
Qualified Charitable Distributions for Still-Working Freelancers: Lowering AGI Without Itemizing (2026)
Compare alternatives
See how CentSense stacks up to other expense and receipt tools for freelancers.
- Keeper Tax alternative
- QuickBooks Self-Employed alternative
- FlyFin alternative
- Expensify alternative
- Shoeboxed alternative
- Veryfi alternative
- Dext alternative
- ReceiptsAI alternative
- Smart Receipts alternative
- EasyExpense alternative
- Zoho Expense alternative
- Rydoo alternative
- Fyle alternative
- Navan alternative
- Expense Tracker 365 alternative
- Paylocity alternative
- Wave Receipts alternative
- QuickBooks Online alternative
- Xero alternative
- See all alternatives โ