Anyone who buys aircraft parts or schedules maintenance work eventually faces the same awkward comparison: the hours or days they expected, and the hours or days that were really spent. This set of questions and answers examines that gap with a cool head. It asks what to record, how to read the numbers, and how to fold the lesson back into the next maintenance estimate without turning it into a bureaucratic exercise.
Why do maintenance estimates drift in the first place?
Because an estimate usually describes the clean version of a job. It assumes the right part is on the shelf, the access panels come off without fuss, and nothing nearby needs attention once the cover is open. Real jobs seldom match that picture. Waiting for a supplier, rechecking a fit, or finding a second small defect all add time that the original guess never included.
That does not mean planners are careless. It means the guess was made from the ideal case, which is the easiest case to imagine. People consistently picture the smooth path first. Recording what actually happened is the only reliable way to see how far from the smooth path a typical job lands.
What exactly should I write down?
Keep the record small enough that you will really maintain it. Four fields are plenty:
- A short description of the task or the parts order.
- The estimate, in hours or days, written before work begins.
- The actual duration, written when the job closes.
- One line on the main reason for any large difference.
Resist the urge to add more columns. A long form discourages use, and the point is to build a habit that survives a busy week.
How do I turn two columns into something useful?
Divide the actual by the estimate for each row. The result is a ratio: a figure of one means the guess was exact, while something near one and a quarter means the job took a quarter longer than planned. A handful of ratios already tells a story. If most sit above one, your estimates lean optimistic. If they scatter widely on both sides of one, the problem is not bias but unpredictability, and the fix is different: widen your range rather than shift it.
The distinction matters. A consistent lean can be corrected with a simple multiplier. A wide scatter calls for splitting the job into smaller parts and estimating each, since small pieces are usually easier to predict than a whole.
Is a small sample trustworthy?
Not very, and it is wise to say so. Three or four entries can mislead, especially if one of them was unusual. A reasonable habit is to hold off on applying any correction until you have somewhere around eight comparable jobs, and to label the factor as provisional until then. Keep jobs of a similar kind together: routine inspections in one group, parts procurement in another, troubleshooting in a third. Mixing them blurs the signal.
When one record is an obvious outlier, such as a job delayed by a shipping problem far outside anyone’s control, mark it but keep it. You may find that the so-called outliers are not so rare after all, which is itself valuable information.
Do parts orders deserve the same treatment?
Yes, and they are often the clearest case. Lead time is an estimate too. A supplier may quote a range, but your own history of how long orders really took is stronger evidence than any quoted figure. Record the date you expected delivery and the date the part truly arrived. Over a season you will have a realistic spread for each supplier or type of part, and you can plan around that instead of the brochure.
It is also worth separating the wait for a part from the time to fit it. They behave differently, and merging them into one number hides which half is causing the surprises.
What does a corrected estimate look like?
Take an invented example. A workshop thinks a particular inspection will take six hours. Its last several comparable jobs took between seven and nine, with a typical ratio of around one and a third. The corrected figure is eight hours, and the planner schedules accordingly. The aircraft is not delayed by the extra two hours because the slot was already long enough. Nobody is rushed, and the work gets the attention it needs.
The same logic applies at home. A short article about tracking where your money actually goes each month uses the identical principle with a household budget: write the guess, check the bill, learn the gap.
Who should keep the record: the planner or the technician?
Ideally both contribute. The person who writes the estimate knows what they assumed. The person doing the job knows what really happened. If only the planner fills in the actual column, details get lost, and the single line explaining the gap tends to become vague. A quick conversation when a job closes, followed by a one-sentence note, captures far more than a report written from memory a month later.
Does this slow the team down?
It adds a minute or two per job, and it saves far more over time. Teams that hold regular, honest comparisons tend to spend less effort on rescheduling, because fewer surprises reach the point where somebody has to apologise to a customer. The cost is mostly discipline. If the record takes longer than a coffee break to update, simplify it until it does not.
One practical tip: put the estimate on the work order itself, so the actual can be added in the same place. Two separate documents drift apart, and a comparison that needs hunting for is a comparison that will not happen.
Where can I read about how other organisations approach learning?
If you like seeing how different fields talk about measurement and improvement, the Cotocus Foundation site belongs to a nonprofit that frames its work around technology, innovation and education, and its menu is split between program pages, research material, a community area and a blog. Aviation is not its subject, so take it as background reading, not as guidance for a hangar.
How often should the numbers be reviewed?
Monthly is a sensible rhythm for most small operations. Look at the newest rows, update the ratio for each job type, and ask whether anything in the working conditions has changed. A new supplier, a new tool or a new team member can all shift the numbers, and the old factor may no longer apply.
Drop entries that no longer reflect how you work. Keeping the history forever feels thorough, but stale data drags the correction in the wrong direction.
What if the estimates still miss after I correct them?
Check whether the misses are random or systematic. If the ratio has simply shrunk, your process improved and the factor should shrink with it. If the misses are still large and scattered, go back to the structure of the estimate. Break it into steps, estimate each step separately, and compare step by step. The step with the widest error is where to focus attention.
Finally, share the numbers with whoever relies on your schedules. Customers and colleagues forgive a longer promised time far more readily than a missed one. A modest estimate backed by your own records builds a kind of trust that few other habits can match.
For a lighter example of comparing a plan against what really happened, a travel piece on a Canadian road trip by car and motorcycle shows how much a route changes once the plan meets the road.

