Maintenance outsourcing: deciding which work leaves the in-house crew

Work that genuinely belongs outside the fence

Three categories move out with little debate. Licensed or statutory activity — pressure systems, lifting examination, electrical work above the site's competence, refrigerant handling — needs a certified party whose competence is auditable. Specialist diagnostics on equipment where the builder holds the software keys or the proprietary alignment rig is a second. Peak load is the third: a shutdown, a relocation or a rebuild demands a headcount no plant carries year-round. Everything beyond those three is a live decision, and the honest question is whether the site intends to be capable on that asset class or is content to buy the capability whenever it needs it.

The knowledge that leaves with the work

Hand a machine's routine maintenance to a contractor and within a couple of years nobody on site can say why it fails, what was changed last time, or whether the settings drifted. The failure history lives in someone else's system, the informal knowledge lives in someone else's technician, and the plant's negotiating position erodes with both. If the work goes out, the records must come back: completed task detail, parts fitted, measurements taken, faults found. Write that into the contract as a deliverable with a defined format, and check it is arriving — an unenforced record clause quietly becomes a blank one.

Contract shapes and the risk they really transfer

Time and materials buys hands and transfers nothing; the site still carries the outcome and the incentive sits the wrong way round. Fixed-scope task contracts buy a defined routine and are easy to audit but do nothing about the failures nobody predicted. Availability or outcome contracts shift some risk to the provider, and price accordingly, but only work where uptime can be measured without dispute and where the provider genuinely controls the variables. A provider who cannot influence how hard operators run the machine will not accept availability risk at a sane price, and should not be asked to.

Controlling a contractor once they are on your floor

The site keeps the duty of care whatever the contract says. That means verified competence rather than a claimed certificate, a proper induction covering site-specific hazards, and permit and isolation control operated by the plant, not the visitor. Somebody on the payroll must own each contract job: confirming the scope, authorising the isolation, accepting the work back and returning the asset to production. Where several contractors work in one area simultaneously, one person coordinates them. Regulators treat a poorly supervised contractor injury as a site failure, and so, in practice, do juries and insurers.

Reading whether the arrangement is working

Cost per asset is the least useful of the available measures because it moves with how much you asked for. Better signals: repeat visits for the same fault, time from callout to a working machine, jobs closed without the required record, and how often the contractor finds work the routine should have caught. Review a sample of completed task sheets rather than the summary invoice. If the same failure keeps recurring and nobody inside the plant can explain the mechanism, the arrangement is buying labour hours and not reliability, whatever the monthly report claims.

Frequently asked questions

Should a maintenance contractor work in our system or their own?

Yours, wherever it can be arranged. Work raised, planned and closed in the site's own records keeps failure history, parts usage and cost in one place and survives a change of provider. Give contractors controlled access rather than letting them run a parallel log. Where their own system is unavoidable — common with equipment builders — insist on a periodic export in an agreed format and load it against the asset, or you will be buying the same history back later.

How do we avoid a machine supplier's service contract becoming the only route to spares?

Deal with it at purchase, not after commissioning. Ask which parts are proprietary, which are catalogue items rebadged, and what the site is permitted to do without voiding support. Negotiate access to diagnostic software, parts lists with original manufacturer references, and training for your own technicians as part of the machine order. Once the asset is installed and running, your bargaining position has gone and the annual service quote reflects that reality precisely.

Is it worth keeping a technician on nights or calling one out instead?

Compare the loaded cost of the shift against the production actually lost while waiting for a callout, on the lines that run at night. Sites often discover the resident technician spends most of the shift on low-value work while the genuinely urgent faults need a specialist who is not there anyway. A middle path — a competent operator trained to clear common stoppages, with an on-call escalation — frequently covers more of the real failure set at lower cost.