How to track engine hours accurately

Engine hours are the foundation of every preventive-maintenance schedule, every cost-per-hour rollup, and every honest answer you give a customer about how much your fleet ran on their job. Get the number right and the rest of the shop floor lines up. Get it wrong — usually by guessing — and the whole PM program drifts into fiction. Here is the daily habit that keeps the number honest, the fallback for when the meter doesn't cooperate, and the four ways crews quietly destroy their own data.

The 60-second daily habit

The job is simpler than most fleets make it. At the end of every shift, the operator of record reads the hour meter on the unit they ran, types the number into the app, and submits. That is the entire workflow. Under good signal it takes about 20 seconds. Under no signal it takes about 30 seconds and queues for sync. Either way, the rule is the same: one log, one operator, one shift, before anyone leaves the yard.

The reason for “before anyone leaves” is not paperwork enthusiasm. It is that human memory of a four-digit number degrades fast. By the next morning the operator will type a plausible value that is plus-or-minus five hours. By Friday they will type a value that is plus-or-minus thirty. Either error is enough to break a 500-hour PM interval if it stacks for a couple of weeks.

Photo-of-meter as the universal fallback

Sometimes the operator cannot type the number. Glove weather, cracked phone screen, app froze, the meter has six digits and the operator only sees the first four through the grime. The universal answer is to take a photo of the meter face.

DirtFleet's hour-log screen accepts a photo in place of a typed number. Our OCR pass reads the digits when it can; when it can't (sunlight glare, fuel-stained glass) the photo sits attached to the log row and a human resolves it in the morning. The point is that the operator never goes home with an undocumented shift. The photo is the receipt.

Two field rules that came from customers, not from us:

• Take the photo with the meter andthe asset number-plate in the same frame when you can. It is the difference between “1,847 hours on something” and “1,847 hours on unit 12.”
• Wipe the glass first. A dusty face costs OCR about 30% of its accuracy on the readings we sampled, and the human-resolution time triples.

What kills accuracy

Four failure modes account for almost every drift case we see in the field. Each one has a fix that does not require buying anything you don't already have.

1. Operator hot-swap during a shift

Two operators share a unit across a single workday — one runs it from 6am to noon, the other from noon to dusk. If only the second operator logs at end-of-shift, the system attributes the whole day to them. Fine for hours math, terrible for safety, certification, and any future incident review.

The fix is a hand-off log at the switchover. Outgoing operator reads the meter, submits, and the incoming operator starts a fresh session from that reading. In DirtFleet the second log auto-fills the “previous reading” field so the delta is computed against the hand-off, not against yesterday.

2. Broken or missing meter

Mechanical hour meters fail. They also get replaced, which resets the count to zero in a way that looks indistinguishable from a stuck reading. The fix is two-part: an asset-level “meter status” toggle (working / broken / replaced) and a manual-entry mode that asks for estimated runtime in hours, not a meter face value. The flag travels with the log so the manager knows which numbers are direct and which are estimated.

Replacement meters get a one-time offset registered against the asset, so the system can keep adding to lifetime hours instead of restarting the PM clock at zero. That offset lives on the Asset row, not in the log, so it survives any future log edits.

3. End-of-week catch-up logging

The most common pattern, and the most expensive. The crew gets through Monday through Thursday without logging because nobody nags them, then on Friday afternoon the shop foreman types in five guesses based on what feels right. The numbers will be directionally correct and arithmetically wrong, every week, forever.

Take a contractor running a CAT 320 excavator at roughly six hours per day. One missed week of logs amounts to 30 hours of estimation drift on a single unit. Spread that across a 15-unit fleet and you are guessing about 450 hours of work per skipped week. The 500-hour PM interval is rounding error to the noise floor of the data.

The fix is a daily nudge — a push notification at the operator's usual end-of-shift time, with one tap to log. Not a daily email to the manager. The manager is not the person who knows what the meter said.

4. “I'll do it later”

The variant of #3 that comes from a well-intentioned operator who plans to log from home, then forgets, then logs the next morning with a guess. Same drift, same cost.

The fix is the same as for any other daily habit: make it easier to do at the moment than to defer. Offline support, a three-tap submit, no required fields beyond the meter reading, and a clear “already logged today” checkmark so the operator can see the streak. Streaks turn out to matter; on our pilot fleets the log-on-time rate went from 71% to 94% in the first month after we added the streak indicator.

How DirtFleet handles each one

• Hot-swap. Hand-off logs are first-class. The incoming operator picks up the outgoing reading without retyping it; the audit trail shows both names.
• Broken meter. Asset has a meter-status field; logs from a broken meter are tagged ESTIMATED and excluded from the cost-per-hour rollup until resolved.
• End-of-week catch-up. Per-operator daily nudge, push to the device that ran the asset last, with a one-tap submit. End-of-week bulk-entry is allowed but flagged BACKDATED so the manager sees what came in fresh versus reconstructed.
• I'll do it later. Offline-first outbox (see Offline-first for the yard) means “no signal” is never an excuse. The log goes into a local queue and ships when the phone next sees a cell tower.

Closing playbook

Print this on a card and tape it to the inside of the cab.

1. Read the meter at end of shift. Before you leave the unit.
2. Type the number into the app. Or take a photo if you can't type.
3. If the meter is broken or replaced, mark it and estimate the hours.
4. If you hand off mid-shift, do a hand-off log so the next operator picks up clean.
5. If you missed a day, log it as backdated. Don't guess silently.

Accurate hours are not a feature. They are a habit, supported by software that makes the habit easier than the workaround. Get the operator-level workflow right and the PM scheduler (see Hour-based PM beats calendar PM) and the cost rollup (see Cost-per-hour is the only fleet metric most owners need) inherit the accuracy for free.

→ Log hours · → Start free trial