An evaluation grid is a measuring instrument, and like any instrument it has a resolution limit. Below that limit, things that exist do not register. On a construction procurement, the things that fail to register are capabilities, and the owner does not learn which ones were missed until the work is underway.
This is usually framed as a bidder’s failing. Write a better proposal, tell the story properly, use the page budget well. All true, and all beside the point for the party who signs the contract. Every capability an evaluation cannot detect is a capability the owner does not get to buy.
The Instrument Measures the File
A grid scores a document. That document is a representation of capability, produced under a page limit, on a deadline, by whoever the firm assigned to write it. The relationship between the quality of the representation and the quality of the builder is real, and it is loose.
Owners read a technical score as a measurement of the firm. It is a measurement of the firm’s file. Those two things correlate well at the extremes and poorly in the middle, and the middle is where most competitive fields sit.
A capability that the grid cannot detect has, in effect, been priced at zero by the owner who wrote the grid.
Four Places Capability Disappears
The losses are not random. They cluster in four predictable places, and all four are visible to an owner who looks for them before the solicitation is issued.
1. Self-performed scope
A firm that subcontracts a specialty trade names the subcontractor, attaches their credentials, and delivers the evidence into the technical envelope as a tidy package. A firm that self-performs the same scope holds that capability inside its own payroll, licences, insurance and crew. It appears in the price envelope as a line item and can be almost invisible in the technical one.
Self-performance is the harder capability to build, the more reliable one to own, and the easier one to score nothing for. Grids that were designed around a subcontracting model systematically undercount it.
2. Projects listed but not scoped
A project list answers the question “have you delivered this type of building.” Modern grids increasingly ask a narrower question: “have you executed this particular scope.” A firm can hold precisely the right experience inside a project already on its list and score nothing for it, because the relevant scope was never drawn out of the project narrative.
3. The page allowance
A page allowance is not a style rule. It is a ceiling on how much evidence the owner has agreed to receive, and material written past it sits outside the evaluation altogether. Submissions routinely spend that allowance according to how important each section feels rather than how much proof it has to carry, which puts the pages in the wrong place.
4. Institutional capability versus individual capability
Grids that score named key individuals capture management depth. They do not capture crew capability. On heritage and specialty work, the decisive competence frequently sits with a tradesperson who will never appear as a named key individual on any submission.
| What the owner intends to buy | What a documentation-weighted grid selects for |
|---|---|
| A firm that has physically executed this assembly before | A firm that can produce evidence it has, in the format requested |
| Craft capability held in-house and controlled directly | Credentials that arrive pre-packaged from a named third party |
| Depth of relevant experience | Depth of experience that fits inside the page allowance |
| The people who will be on the site | The people the grid defined as key individuals |
Why This Belongs to the Owner
A grid does not merely measure a field. Over time it shapes one. Firms optimise for what scores, because firms that do not optimise for what scores stop winning work. A grid that consistently rewards documentation quality will, over several cycles, select for firms that are good at documentation.
That is a tolerable outcome on commodity work. On heritage rehabilitation and specialty institutional work it is expensive, because the capabilities hardest to document are frequently the ones that decide whether the work is any good: a craftsman who has done this assembly, a crew that has opened up this kind of wall before, judgement that lives in people rather than in a process manual.
The owner is not a passive recipient of whatever quality the market happens to offer. The owner designs the instrument that decides which parts of the market become visible.
Designing a Grid That Can See
None of this requires abandoning a structured evaluation, and none of it requires trusting anyone. It requires asking for evidence in a form that capability can actually take.
Five Corrections
Companion piece
This argument sits alongside The Cheapest Fee Buys the Most Expensive Building, which examines how price weighting decides outcomes that owners believe they are deciding on merit. The two failures compound: a grid that cannot see capability, scored against a price line that can always overturn it.
The Fiduciary Reading
Procurement documents are usually treated as administrative instruments and drafted accordingly, often by people whose duty runs to process compliance rather than to the owner’s balance sheet. They are not administrative instruments. They are the specification for how the owner will perceive the market, written before the owner has seen it.
An owner who awards to the best-documented firm has not necessarily awarded to the best-qualified one. Sometimes those are the same firm. When they are not, the difference does not disappear. It reappears during construction, in the assemblies that were harder than they looked, executed by people who had not done them before.
The correction is available and it is cheap, because it happens entirely on paper before anyone has mobilised. Write the grid so that the capability you actually want is a capability your grid can find.