Choosing the Right MEP Contractor for a High-End Project

On a high-end project, the MEP package is where the money, the risk and the eventual owner experience concentrate — and yet it is usually procured on a comparison of numbers attached to non-comparable scopes. Having sat on every side of this table — designing, tendering, delivering, and being brought in to rescue — this is the vetting framework we would use if we were the client.

First, decide what you are actually buying

MEP firms sit on a spectrum. At one end, labour-and-materials installers: competent hands that build what drawings tell them, and only what drawings tell them. At the other, design-and-build engineering firms that produce the calculations, models and coordinated drawings, then install and commission from them. Neither is wrong — but a project without a complete, coordinated MEP design cannot safely buy the first kind, because the design gaps will be resolved ad hoc on site, priced as variations, and owned by nobody. Most high-end residential disputes we are asked to untangle began exactly there.

The seven questions that expose the difference

One: who does your design — in-house engineers or a subcontracted consultancy — and can we meet them? Two: will the services be modelled and clash-checked before installation, and to what LOD? (Our guide to BIM coordination explains what good answers sound like.) Three: show us a commissioning report and an O&M manual from your last comparable project — the documents reveal the culture. Four: who exactly will run our job, and what else will they be running? Five: what is your directly-employed labour versus subcontract ratio for each trade? Six: which manufacturers accredit you, and who honours the warranty in year four? Seven: walk us through a variation you raised on a recent project and how it was valued — the answer tells you how change will feel for two years.

Red flags with excellent camouflage

The suspiciously low tender that has quietly excluded controls, commissioning, attenuation, builders' work or certification — the gap reappears later as variations, with leverage reversed. Provisional sums covering half the technology scope. A portfolio of photographs but no reference clients you may actually telephone. Design responsibility accepted in the contract but no engineers on the payroll. Glossy proposals that cannot produce a sample calculation, drawing or commissioning sheet. And programme promises with no resource plan behind them — optimism is not a methodology.

Weighing price against the whole life of the building

The tender figure is one number in a longer equation that includes the variations the contractor's design maturity will or won't generate, the programme risk their coordination discipline will or won't create, the energy and maintenance costs their equipment selections lock in for fifteen years, and the value of documentation when you refinance, sell, or extend. A contractor 8% higher who models the building, commissions rigorously and hands over real documentation is very frequently the cheaper contractor measured over five years. That is not a self-serving claim — it is the consistent arithmetic of the rescue projects we inherit.

Structuring the engagement for success

Whatever firm you choose: appoint early — the biggest MEP savings are made at design stage, not at tender. Make deliverables contractual: coordinated models, commissioning results against stated design criteria, as-builts, O&Ms, training, and — for automation — source code and configuration files. Hold a sensible retention against seasonal proving. And insist on named people, because you are not hiring a logo; you are hiring the project manager, the design engineer and the commissioning lead who will actually answer the phone.

Where AVC deliberately sits

We built AVC as the firm this article describes: engineers who model what they install, install what they modelled, commission against their own calculations, and document the result — across MEP and the technology layer that other contractors treat as someone else's problem. Use the seven questions on us too. That is rather the point of publishing them.