Specification-driven cost: the requirement decides the price before the tender does
A specification is a commercial decision, and it is usually made by people who are not in procurement and are not measuring cost. Its effect is felt twice: once as a premium on the item, and once — normally larger — as the suppliers eliminated before any tender exists. A parameter that only one manufacturer can meet has set the price irrespective of how the negotiation is run. The method is to separate the requirement into function, performance and prescription, trace where each parameter came from, and count how many qualified suppliers remain at each tolerance level.
- Applies to
- Strategic categories primarily, and any category where the requirement is written before the market is consulted
- Unit of analysis
- One specification or scope of work, before it is issued
- Position in the lifecycle
- Before the tender, which is the only point at which this is cheap
The requirement is a commercial document that nobody prices
By the time a specification reaches procurement it usually reads as a technical fact. It is not. Every parameter in it is a decision that has narrowed the field of possible suppliers and set a floor under the price, and most of those decisions were taken without anyone calculating what they cost.
This is not a failure of engineering judgement. The people writing a specification are optimising for something real — performance certainty, compatibility with an installed base, avoidance of a failure they have personally dealt with. They are not measuring the commercial consequence because it is not visible to them, and it is not visible because it materialises later, in a market they do not interact with.
The consequence is that the most expensive procurement decision in a category is frequently taken before procurement is involved, and taken by someone with no reason to know they were taking it.
Over-specification costs twice, and the second cost is larger
The familiar version of this argument is the premium: a tighter tolerance, a higher grade, a longer warranty each carry a unit cost, and relaxing them recovers it. That is true and it is the smaller half.
The larger half is competitive. Each parameter eliminates a portion of the supplier field, and the eliminations compound. A specification that four suppliers can meet produces a different price from one that fourteen can meet, and the difference is not a function of how well the tender is run. It has already been decided.
The evidence for the magnitude of that second effect comes from public procurement, where competition is recorded and the outcomes are published. The OECD finds single bidding associated with prices higher by 9.6%. The European Court of Auditors records the EU single-bid rate rising from 23.5% in 2011 to 41.8% in 2021, with competition decreasing across the decade despite directives designed to increase it. Peer-reviewed analysis of more than 17,000 Finnish invitations to tender found a median of two bidders, nearly a third of tenders attracting none, and — the finding that matters here — a price effect of competition that is larger in low-competition environments.
That last point inverts the usual intuition about where to spend effort. The marginal value of restoring one credible supplier is highest exactly where the field has already been narrowed to two or three, which is to say in precisely the categories where a specification has done its work.
Relaxing a parameter can move the category between Kraljic quadrants
The consequence of the above is that specification is not only a cost lever. It is one of the few instruments that changes a category’s structural position.
A bottleneck category — high supply risk, modest spend — is frequently bottleneck because of a parameter. Where the constraint is a tolerance, a material grade or a certification that one supplier holds, relaxing it converts the category to leverage, and the entire posture changes with it: competition becomes available, dual sourcing becomes possible, and the negotiation stops being a conversation about goodwill. The procurement iceberg and the Kraljic matrix sets out where hidden cost sits in each quadrant.
This is worth stating because it reframes what the exercise is for. Specification review is usually justified as a cost-reduction activity and ranked accordingly against other cost-reduction activities. Its more valuable output is optionality, and optionality does not show up in a savings report.
Trace each parameter to its origin, then count the field
Step 1 — Separate function, performance and prescription
Most specifications mix three kinds of statement, and only the first two carry a requirement.
Three kinds of statement in a specification, and what each one costs| Type | What it states | Example | Commercially |
|---|
| Functional | What the item must achieve | "Must isolate the circuit within 40 ms" | Load-bearing. Rarely negotiable |
| Performance | The measurable threshold that constitutes achievement | "Operating temperature −20 °C to +65 °C" | Load-bearing, but the threshold itself is a choice |
| Prescriptive | How it must be built, or which part must be used | "Grade 316L stainless; manufacturer part number XYZ" | Frequently not a requirement at all — it is one solution recorded as though it were the need |
Prescriptive statements are where the field narrows fastest and where the least is lost by challenge. A part number in a specification eliminates every supplier who could meet the function by another route, and it does so silently: the tender returns a small field and nobody attributes that to the document.
The rewrite that follows is not remove the parameter. It is restate the parameter at the level of the need it was standing in for.
Step 2 — Trace where each parameter came from
Every parameter has an origin, and the origin determines whether it can move. Four cases, of which only two are binding.
Parameter provenance and whether it binds| Origin | How it presents | Binding? |
|---|
| Functional necessity | The item does not work without it | Yes |
| Regulatory or certification requirement | Mandated by a standard, an approval or a customer’s own compliance obligation | Yes, and it must be verified rather than assumed — a surprising share of these turn out to be an internal interpretation of a standard rather than the standard |
| Legacy carry-over | Present because it was present last time. Nobody can say who set it | No |
| Stakeholder preference | A named person is comfortable with it. Often based on one prior failure | No, but it requires a conversation rather than a redline |
The diagnostic question for each parameter is simply: what happens if this is not met? A parameter whose owner cannot answer that in concrete terms is in one of the bottom two rows.
The second row deserves particular attention because it is where the largest recoverable cost usually sits. Regulatory over-interpretation — applying a standard more strictly than the standard requires, or applying a standard intended for one application to another — is common, defensible-sounding, and expensive. It is also checkable against the text of the standard, which makes it one of the few specification arguments that can be settled rather than negotiated.
Step 3 — Count the supplier field at each tolerance level
This is the step that distinguishes the method from a general encouragement to challenge requirements, and it is the one that produces a number.
For each candidate parameter, establish how many qualified suppliers remain if the parameter is held, and how many if it is relaxed to the next defensible level. The output is a sensitivity ranking: which single parameter is costing the most competition.
Illustrative — supplier field by parameter. The value of the table is the ordering, not the counts.| Parameter | Held as written | Relaxed to next defensible level | Field change |
|---|
| Material grade | 3 suppliers | 11 suppliers | +8 |
| Dimensional tolerance | 9 suppliers | 12 suppliers | +3 |
| Certification scope | 4 suppliers | 4 suppliers | 0 — regulatory, binding |
| Lead time commitment | 6 suppliers | 10 suppliers | +4 |
Illustrative figures. The table identifies which conversation to have first, and it converts an argument about engineering preference into an argument about how many companies are allowed to quote.
A parameter with a field change of zero is settled and should be conceded early. Conceding it visibly is what buys the credibility to challenge the one above it.
Step 4 — Price the change on both axes, and record what is given up
Each proposed change carries a unit cost effect and a competition effect, and they are not the same number. A relaxation that saves nothing per unit but doubles the field may be worth more than one that saves 4% per unit and changes nothing structurally.
Each change also carries a risk, and the risk register is what makes the proposal survivable in front of the stakeholder who wrote the parameter. For each change: what quality, compliance or performance exposure is accepted, who accepts it, and what would have to be true for the decision to be wrong. A specification challenge presented without that register reads as procurement trying to save money at engineering’s expense, and it will be resisted on those terms regardless of the arithmetic.
For services, the failure runs in the opposite direction
Goods specifications fail by being too precise. Service specifications fail by being too vague, and the cost arrives later rather than at award.
An ambiguous scope of work produces a low bid and an expensive delivery, because everything the document did not settle becomes a change request at the supplier’s rates rather than the market’s. The mechanism is the mirror image of over-specification: under-specification also removes competition, but it removes it after the award, when the buyer has no field left to appeal to.
The parameters that matter in a scope of work are therefore the ones that determine what is included:
- Deliverable definitions with acceptance criteria.
- The change control procedure, and who may invoke it.
- The precedence order between documents.
- Service levels with a measurement method attached.
- Remedies that have a financial consequence. A service level without a remedy is a statement of intent.
The same four-step logic applies, with the field test replaced by a completeness test: for each obligation, can a supplier who wanted to argue about it find room to do so.
What this method does not do
- It does not overrule engineering. It produces a costed set of options and a risk register; the decision on each parameter belongs to whoever owns the consequence of it being wrong.
- It does not work retrospectively. Once the tender is issued, the field is set and the specification’s cost has been incurred. Reopening a specification mid-process carries a schedule cost that usually exceeds the recovery, which is why this is a pre-tender exercise and not a post-award one.
- It does not establish the right price. It establishes how much competition the requirement permits, which is an input to price rather than a substitute for cost analysis.
- It has no continuous component of its own. A specification is written, reviewed and issued. What changes continuously is the market that will respond to it, which is a different question addressed by market intelligence rather than by monitoring the document.
The competitive mechanism is sourced; the premium figures are not
The competitive mechanism is documented. The OECD single-bidding finding, the European Court of Auditors trend data and the Finnish tender analysis are all public procurement, all primary, and all carry the transfer caveat stated above.
The hidden-cost literature is settled and old. Ellram’s work on total cost of ownership establishes the classification of cost elements beyond purchase price, and Ferrin and Plank extend it to technology procurement, identifying the training, integration and organisational-change categories that a specification implicitly commits to. Both are peer-reviewed and named. The finding that firms understand these concepts and largely do not apply them dates from 1993 and has not been overturned.
Where this runs in EXOS
Specification work is a chain because the requirement, its cost and its enforceability are three separate analyses, and they are usually performed by three different people.
The scenario chain that executes this method| Step | Scenario | What it produces |
|---|
| Separate the requirement from the wish list | S13 · Requirements Gathering | MoSCoW prioritisation of raw stakeholder requirements, dependency map identifying conflicts, constraints and assumptions register recording what is explicitly out of scope and why |
| Identify over-specification and price the alternatives | S8 · Specification Optimisation | Gold plating report with cost impact per parameter, comparison against industry-standard tolerances, alternative specification maintaining function at lower cost, supplier market expansion showing how each change widens the qualifying field, and a specification change risk register stating what is gained and lost per parameter |
| Quantify what a parameter is worth | S2 · Cost Breakdown / Should-Cost | Component-level cost structure showing which parameter drives which cost line, and the negotiation headroom that follows |
| For services: test whether the scope is enforceable | S16 · SOW Critic | Clause-by-clause findings by severity, contract protection score across enforceability and completeness, questions to resolve before signature, and replacement clause language for each gap |
The supplier-market-expansion output of S8 is the one to lead with in an internal conversation. It reframes the discussion from procurement wants a cheaper part to this parameter is the reason three companies can quote instead of eleven, which is a different argument and a harder one to refuse.
Upstream input — Market Snapshot (S28). The field counts in step 3 have to come from somewhere. A market snapshot on the specific niche — not the category, the niche — establishes what tolerance levels the market actually offers and which suppliers exist at each. Query specificity determines whether this returns intelligence or a summary of publicly available material.
Downstream — S9 · RFP Generator takes the revised specification into a tender package with evaluation criteria attached.