Skip to content
allanninal.dev/build
Collaborate

Part 3 of 7 · Equipment maintenance scheduler series ~6 min read

How a deferral gets recorded

This is the post that matters. A scheduler that only knows done and overdue will show an overdue list of forty items within six months, at which point it has stopped being information.

Key takeaways

  • Deferral is a button, not a failure. Without one, jobs go overdue and stay there.
  • A deferral needs a reason, a name and a new date. All three.
  • The second deferral of the same job is a different event and is shown as one.
  • Percentage past interval is more useful than days overdue.
  • Report the total deferred, not just the count. It is a debt.

The deferral record

How a maintenance job deferral is recordedA vertical chain of five steps entered by a box labelled A due job that cannot be done now. Step one asks whether it is statutory, checked first; if so it exits to No deferral available, which is a different conversation. Step two records a reason from a short list plus free text. Step three requires a new date, not simply later. Step four records whose decision it was, a named person. Step five shows how many times it has now been deferred, with a side box noting that a third deferral escalates. A note says requiring a date is what stops deferral becoming an indefinite postponement.AWS ACCOUNTA due jobcannot be done nowIs it statutory?checked firstNo deferral availablea different conversationyesReasonfrom a short list,plus free textNew datenot 'later'Whose decision?a named personHow many times now?shown on the recordThird deferralescalatesRequiring a date is what stops deferral becoming an indefinite postponement.
Fig 1. What a deferral requires. Each field removes a way for a deferred job to disappear.
  • App integration
  • Security & identity
  • Management
  • Analytics
  • People

Reasons from a short list

Five or six options cover almost everything: part not available, machine in use, no engineer available, weather, waiting on a third party, deprioritised. A short list makes the aggregate analysis possible, and the free text next to it captures the specifics.

The distribution of those reasons over a year is one of the more useful outputs of the whole system. If half of all deferrals are parts availability, the fix is a stock policy rather than a scheduling change, and that is not obvious from any individual deferral.

Percentage, not days

Three overdue maintenance jobs measured as a percentage of intervalA bar chart with three bars showing the percentage of the service interval used. A pump fourteen days over is at one hundred and four per cent. A press fourteen days over is at one hundred and thirty-one per cent. A chiller forty days over is at one hundred and eight per cent. A note says days overdue ranks these wrongly, and the press is the urgent one at one hundred and thirty-one per cent.050100150200~104Pump, 14 days over~131Press, 14 days over~108Chiller, 40 days overPercentage of interval usedDays overdue ranks these wrongly. The press is the urgent one, at 131%.
Fig 2. Three overdue jobs measured two ways. Days overdue puts the chiller at the top; percentage of interval used correctly puts the press there.

The press is fourteen days over and thirty-one per cent past its interval because it is in heavy use; the chiller is forty days over and barely past interval because it has hardly run. Sorting the overdue list by days produces exactly the wrong order of work.

This is a small change to a sort key and it changes which machine gets attention on a busy Tuesday, which over a year is most of what a maintenance system is for.

The second and third deferral

What changes each time

  • First deferral: recorded, shown as deferred rather than overdue, no fuss.
  • Second deferral: the record shows both reasons and the original due date. Still routine.
  • Third deferral: goes to whoever owns the equipment, not just the maintenance list, with the full history.
  • Beyond that: the job appears on the monthly report by name, every month, until it is closed.
  • Nothing is ever blocked. The escalation is visibility, not permission.
  • Because the alternative is that people stop recording deferrals and go back to silently skipping.

That last point is the design principle behind the whole feature. Any mechanism that makes deferring painful will be routed around, and a routed-around deferral is invisible, which is the situation the system exists to prevent. Escalating visibility is the only pressure that does not create an incentive to hide.

Deferral is not always wrong

A machine that is about to be replaced, a service that would displace a job with a genuine deadline, a part that will arrive in a week — deferring in those cases is correct, and a system that treats every deferral as a failure is asking people to lie about their reasoning.

What matters is that somebody looked at it and decided, and that the decision is attached to a name. The failure mode is not deferral; it is drift.

The debt

How deferred maintenance is reported as an accumulated debtA horizontal row of five boxes. Nine jobs deferred this quarter. Not a count, a total. Thirty-seven machine-weeks past interval. Trending up, from twenty-two last quarter. That is the number to put in front of people. A note says nine deferrals sounds manageable and thirty-seven machine-weeks does not.DEFERRAL AS A DEBT9 jobs deferredthis quarterNot a counta total37 machine-weekspast intervalTrending upfrom 22 last quarterThat is the numberto put in front of peopleNine deferrals sounds manageable. Thirty-seven machine-weeks does not.
Fig 3. How deferrals aggregate. Expressing the total as accumulated time past interval turns a list of individually reasonable decisions into a number somebody has to answer for.

The trend is the part that produces action. Any single quarter’s deferrals look defensible, and a debt that has grown from twenty-two to thirty-seven machine-weeks in three months is a resourcing conversation with evidence in it.

Next: the jobs that cannot be deferred at all.

All posts