Why end-of-month estimates break PM
End-of-month "let me just guess what the meter said" is the most common reason PM intervals drift. Here's the math of why a 5% estimate error breaks a 500-hour service schedule.
A hydraulic pump on a wheel loader died at 540 hours. The 500-hour service had been done at what the spreadsheet called 480 hours. The actual reading on the meter when the tech opened the panel was 612. Someone in the office had been writing end-of-month estimates for nine months. The repair cost $14,200. The PM that would have caught it would have cost $340. This post is about the math of why “let me just guess what the meter said” is the single most expensive habit on a construction fleet.
Why estimates compound
An honest estimate of a single shift is plus-or-minus an hour. It feels harmless. Eight hours on a Tuesday, six on a Wednesday, eight on a Thursday — the foreman pencils in 22 hours for the week and moves on. The reality might be 23.5, or it might be 20.5. Call it a 5% error. Inside one week it is noise.
The trouble is that the error has no correcting force. Nothing in the workflow pulls the estimate back toward the real meter reading. The next month, the foreman pencils another estimate from the last estimate. The month after that, another. The bias is whatever direction the foreman happens to guess; the magnitude grows linearly because each guess is on top of the last.
By month nine, the spreadsheet and the meter disagree by an amount large enough to skip a PM interval entirely. The pump does not care which number was in the spreadsheet.
The 5% rule
A simple lens that fleet managers can carry in their head: a consistent 5% logging error on a unit running 2,100 hours per year (roughly six hours per day, five days a week, fifty weeks) equals 105 hours of annual drift. That is enough to skip the 500-hour PM by 20%, or skip a 1,000-hour interval by 10%. Neither is a small number when the failure mode is a pump or an injector.
What different error rates cost you
Below is what the annual drift looks like for the same unit — a wheel loader running 2,100 engine hours a year — at three different sustained error rates. The compounding here is not about each month being multiplied by the last; it is about the drift summing month after month because no real meter reading ever pulls the spreadsheet back to the truth.
- 1% sustained error — about 21 hours of drift per year. Inside the safety margin of most PM intervals; you probably never notice it.
- 5% sustained error— about 105 hours of drift per year. Enough to skip a 500-hour PM by a full month's work, or arrive at a 1,000-hour PM 10% late. This is the most common pattern we measure on fleets that do month-end estimates.
- 10% sustained error— about 210 hours of drift per year. Enough to skip a 1,000-hour interval by more than 20%, and to make every cost-per-hour rollup unreliable. Common on fleets that estimate quarterly or rely on the foreman's memory.
What the drift actually costs
The cost of a skipped PM is not the cost of the PM you skipped — it is the cost of the failure the PM was designed to prevent. Three buckets, all of them real numbers our customers have shown us:
- PM warranty void.OEM warranties on drivetrains and hydraulic systems require documented service at the interval the OEM specified. “Our spreadsheet said we were on time” is not a defense the claims adjuster accepts when the meter on the unit shows otherwise. One denied claim on a major system is a five-figure event.
- Surprise breakdowns. A 320-class excavator down for two days during a tight schedule is a delayed pour, a missed pump truck, and a foreman who has to find another machine. The downtime cost is typically 5–10x the cost of the part that failed.
- Accounting mismatch with payroll hours. Operators clock their hours through payroll. Their machines should be running for roughly that long. When the asset hours and the payroll hours diverge by 105 hours a year per unit, the bookkeeper starts asking questions about either ghost overtime or unaccounted-for equipment usage. Neither is a conversation a fleet manager wants to have without data.
How daily logging eliminates the problem
The fix is not better estimates. It is a shorter feedback loop. A daily reading puts a real number against a real meter every 24 hours. The error on any single log is plus-or-minus zero — you read what the meter says. The error on the rollup is the sum of zero errors, which is zero.
The objection to daily logging is always the same: the operators won't do it. The answer is to make it easier than the workaround. Three taps in an app on the device that already lives in the cab. Offline support so signal is not an excuse. A photo-of-meter fallback for the days when the operator can't type. A daily nudge so the habit lands at the same time every shift. None of this is novel; most fleets we've worked with raise their log-on-time rate from roughly 70% to roughly 95% in the first month after the pieces are in place.
Once the daily log is steady, the PM scheduler (see Hour-based PM beats calendar PM) operates on real meter readings. The cost-per-hour rollup (see Cost-per-hour is the only fleet metric most owners need) operates on real hours. The whole stack starts agreeing with itself.
A manager-readable summary
For the conversation with the owner or the GC:
- End-of-month estimates introduce a sustained 5–10% error in machine hours.
- That error sums into 100–200 hours of annual drift per unit on a normal-utilization fleet.
- On a 500-hour PM interval, that drift is enough to skip service entirely once or twice a year per unit.
- The cost of one skipped PM that turns into a failure ($14,200 in the loader story above) pays for the software for multiple years on a 25-asset fleet.
- Daily logging — three taps, on the device that's already in the cab — eliminates the entire failure chain. See our pricing for what daily logging costs at the fleet level.