Don't Backfill Utilisation From Ignition-Only
When engine hours are missing or delayed, do not invent utilisation from ignition on/off events alone. Keep ignition as an event stream, leave the hours gap visible, and wait for a locked hours source.
Imagine a plant or fleet lead staring at a blank hours cell for last week while the ignition log looks tidy. Someone says: “Count the time between ignition-on and ignition-off — that is close enough for utilisation.” That step invents an hours series the locked source never produced. It is a different failure from treating a missing or delayed hours reading as zero, and from validating engine hours versus powered-on time on one sample. This article is a practical office and field method for that moment: refuse ignition-only backfill, keep both streams named, and pause utilisation until hours catch up or the source is re-locked. It also complements the rule that GPS movement is not engine hours — another way offices invent work from the wrong signal.

Illustration for this article; it does not show a customer deployment, measured result or certified installation.
Ignition is an event stream, not an hours source
Ignition on and ignition off tell you that a power or key circuit changed state. They do not, by themselves, produce the same series as a cab hour meter or a locked telematics engine-hours feed. Treating the gaps between ignition events as “hours worked” quietly swaps the utilisation file onto a different source without naming the change.
That swap is easy to miss in a weekly report. The sheet still has one number per machine. The unit still says “hours.” Only the provenance changed — from a locked hours reading to an inferred interval. Keep the labels honest: ignition events stay ignition; hours stay hours.
What ignition-only can and cannot support
Ignition logs are useful. They are not a substitute hours source for utilisation comparisons.
| Situation | What it can support | What it does not mean |
|---|---|---|
| Hours cell empty; ignition on/off present | Note that the machine showed power-state changes; escalate the missing hours feed | That utilisation for the week can be calculated from those events |
| Hours delayed; ignition looks complete | Keep the period open; compare when hours arrive | That ignition intervals are “close enough” to publish as hours |
| Long ignition-on with little movement | A flag for idle or powered-on review — separately from hours | That GPS stillness or travel fills the hours gap (see related GPS rule) |
| Ignition events after a GPS gap | Evidence that power state resumed; still not a work-hours total | That the gap can be backfilled as productive engine hours |
Do not backfill utilisation from ignition-only
Backfill here means writing a utilisation or hours total into the file using only ignition intervals, then treating that total as if it came from the locked hours source. Refuse that step even when the report deadline is close.
- Do not sum ignition-on to ignition-off and paste the sum into the hours column.
- Do not invent “estimated hours” from ignition density and then rank machines by that estimate.
- Do not average an ignition-derived figure with a partial hours reading — that stacks two refusal rules (see do not average two disagreeing hours sources).
- Do keep the hours cell blank or marked unresolved until a locked hours reading exists for the window.
- Do store ignition events in their own named field if the office needs them for ops review.
Keep ignition and hours as separate named streams
Offices get into trouble when one column silently serves two meanings. Use a short written rule for the file:
- Name the locked hours source (cab meter, telematics engine hours, or another source the team already locked).
- Name ignition (or powered-on events) as a separate stream — never as the utilisation total.
- When hours are missing or delayed, leave utilisation unresolved for that machine and window.
- When hours return, write the locked reading; do not “reconcile” by blending ignition intervals into the same cell.
- If the team later decides to change the locked source, treat that as a source change — not a continuous record (see change of hours source is not one continuous record).
What you can and cannot conclude
| You can conclude | You cannot conclude |
|---|---|
| The hours feed did not deliver a reading for this window | That utilisation was zero, full, or “about N hours” from ignition alone |
| Ignition events occurred and can be reviewed as events | That those events equal engine hours or productive utilisation |
| Comparisons across machines should wait for locked hours | That ranking by ignition intervals is a fair utilisation leaderboard |
| A sample machine can validate hours vs powered-on before rollout | That a fleet-wide ignition proxy replaces sample validation |
Pause rules before utilisation comparisons
Before you compare utilisation week-over-week or machine-to-machine:
- Confirm every row in the comparison uses the same locked hours source.
- Exclude or flag rows whose hours cell was empty and later filled from ignition-only.
- Do not close a gap after a missing window by spreading ignition time into neighbouring days — that mirrors the refusal for a sudden hours jump after a gap.
- If only ignition is available, report operational notes separately; do not publish a utilisation KPI from that stream.
Frequently asked questions
If the hours feed is delayed but ignition events look complete, can we estimate utilisation from ignition?
No. Ignition on/off is an event stream, not an engine-hours reading. Leave the hours gap visible, keep ignition as ignition, and wait for the locked hours source before writing utilisation.
Is this the same as engine hours vs powered-on time?
Related but not the same. Engine hours vs powered-on time is about validating which signal you locked. This article covers a different failure: inventing utilisation from ignition events when the hours series is missing or delayed.
What if the weekly report needs a number before hours catch up?
Mark the period unresolved and keep raw ignition events separate from hours. Do not publish a backfilled utilisation total as if it came from the locked hours source.
Does this replace the rule that a missing hours reading is not zero?
No. Missing still is not zero. This page adds the next refusal: do not replace the missing hours cell with an ignition-only estimate either.