Rolling a Maintenance Programme Across Sites

A maintenance programme designed centrally and announced to sites produces compliance for six weeks and then quietly stops. The design is rarely the problem.

Sequence

One: asset audit. Everything else references it. A day per site, and it cannot be done remotely.

Two: vendor consolidation. Agreements in place before scheduled work begins, or the schedule has nobody to execute it.

Three: pilot at three to five sites. Different formats, different ages, different regions. Long enough to include a full preventive cycle and at least one real failure — a quarter minimum.

Four: revise. The pilot will find that the task lists are wrong, the reporting form asks the wrong questions, and site managers do not do the thing you assumed was obvious.

Five: roll out in waves, by region, so the central team can support each wave.

Six: sustain, which is the part that fails.

The pilot has to be allowed to fail

A pilot designed to succeed teaches nothing. Choose sites that include the difficult cases: the oldest kitchen, the site with a weak manager, the one furthest from any vendor.

Measure what actually happened, not whether people liked it. Completion rates for site tasks, work order quality, vendor response, whether the asset register survived contact with reality.

Expect the site checklist to be the failure point. Manager-completed tasks have the lowest compliance of any component, and finding out at five sites is much cheaper than at fifty.

What site managers need

They did not ask for this and it is added to a job that is already full.

Explain the why once, concretely. Not "asset management" — "the walk-in at site 12 failed in July, we lost eleven thousand dollars of product, and the condenser had not been cleaned in two years."

Make reporting faster than the current method. If the new process is slower than calling the contractor directly, they will call the contractor directly. This is the single most common cause of rollout failure and it is entirely a design problem.

Keep the monthly checklist to ten items.

Give them something back. Faster response, fewer failures, less time on the phone. A programme that only extracts information from sites will be resisted.

Do not make them diagnose. Ask what it is doing, not what is wrong.

What the central function needs

One owner. Not a committee and not a shared responsibility. A named person whose job this is.

Authority over vendor selection and spend thresholds, or the programme becomes advisory.

Access to the sites. Somebody has to visit, and remote programmes drift.

A small reporting set that goes to operations leadership monthly, so the programme has visibility above the person running it.

Where rollouts fail

No asset register, so the software is empty and everything references nothing.

Site tasks nobody does, because they are too long, unexplained, or not checked.

Vendors who ignore the process and continue as before. This requires the agreement to specify it and someone to enforce it.

A reporting burden with no feedback. Sites submit data and never see anything come back.

The champion leaves and nobody owns it.

Budget cut in the second year, after the failures have stopped and the value has become invisible.

Sustaining it

Monthly: completion rates by site and vendor. Open work orders ageing. Emergency versus scheduled proportion.

Quarterly: cost per site, repeat failures, vendor performance review.

Annually: schedule revision against work order history, replacement planning, vendor re-tender or renegotiation.

Visit sites. Look at a condenser that was cleaned last week. The programme is real to the extent that someone checks.

Report the avoided failures, even though they are invisible. "Refrigeration emergency call-outs are down from thirty-one to nine year on year" is the sentence that protects the budget, and it requires having measured the before.

The realistic timeline

Audit and vendor work: a quarter. Pilot: a quarter. Rollout: two to three quarters depending on estate size. First full year of clean data: the year after that.

Anyone promising results in a quarter is describing software installation, not a maintenance programme.

The pilot report

What to write at the end of the pilot, before rolling out. One page.

Completion rates. Site checklists, preventive tasks, work order close-outs. By site.

What broke in the process, specifically. Which fields nobody filled, which steps were bypassed, which vendors ignored the requirement.

What site managers said, including the complaints. The complaint that the process is slower than calling the contractor is the most important finding you will get.

What the asset register got wrong. It will have errors and they need correcting before the pattern is replicated.

What it cost, in central time as well as vendor spend.

What changes before wave one.

A pilot that produces no changes was not a pilot. If the report says everything worked, either the sites were chosen to succeed or nobody asked properly.