The cheapest bid is often the most expensive: why every buyer needs a Should Cost Model
Every experienced procurement professional has watched it happen. A bid comes in dramatically below the field, the evaluation model rewards it, the contract is signed, and within a year the savings have evaporated into change requests, service failures or a supplier quietly renegotiating from a position of strength. The Cabinet Office's Should Cost Modelling guidance, published in May 2021, gives this failure a name, low-cost bid bias, and a defence: know what the thing should cost before anyone tells you what they will charge.
Low-cost bid bias is a system failure, not a personal one
It is easy to blame evaluators for chasing cheap bids, but the bias is structural. Price is the easiest criterion to score, budgets are tight, and without an independent reference point the lowest number looks like the best value. The guidance's answer is to manufacture that reference point in advance: a whole life forecast, built from real data on day rates, volumes and assets, with uncertainty modelled honestly rather than hidden.
The referral rule gives the model teeth. A bid more than 10 percent below the average of the others, or below the Should Cost estimate, must be referred before acceptance. That single rule changes evaluation room dynamics. The question stops being can we justify taking the cheap bid and becomes can the bidder explain, line by line, how they arrived at a number the evidence says is undeliverable.
Whole life is the phrase that changes decisions
The guidance is emphatic that the key factor is whole life cost, not initial purchase price. For construction that means modelling operation and maintenance across the design life, not just the build. For services it means comparing genuine delivery options, in-house, market, or a mixed economy, on the same whole life basis, so the delivery model assessment rests on evidence rather than ideology in either direction.
This is also where models earn their keep after award. A model that forecast the whole life cost becomes the baseline for open book contract management, budget setting and variance analysis. The organisations that get the most from Should Cost Models are the ones that keep them alive as performance tools, updated with actual costs, rather than filing them with the business case.
A model without governance is just a spreadsheet with opinions
The least glamorous parts of the guidance are the most important: the Initial Model Assessment, the written specification, the Book of Assumptions, the separation between the people who build the model and the people who test it. Cost models influence decisions worth millions, and an unreviewed formula error or an undocumented assumption can distort a delivery model choice or an evaluation. Proportionality cuts both ways: a simple requirement deserves a simple model, but a critical decision deserves formal QA, named roles and a Model SRO who owns the result.
Data discipline is the other half. Assumptions must be logged, sources recorded, gaps acknowledged. A model whose inputs cannot be traced convinces nobody, least of all a bidder being challenged over an abnormally low price.
Start earlier than feels natural
The guidance asks for the model to be agreed at planning stage, before procurement documents are published, and for good reason. Built early, the model shapes the requirement definition, the procurement timetable and the disclosure decision, whether bidders will see the model, which determines whether it can be used in formal evaluation at all. Built late, it becomes a retrofitted justification for decisions already taken.
So treat the Should Cost Model as the first commercial artefact of a procurement, not the last analytical one. The teams that do consistently report the same experience: the model pays for itself the first time it catches a bid that was too good to be true.
The takeaways
- Low-cost bid bias is structural: an independent whole life forecast is the defence.
- Bids more than 10 percent below the field or the model estimate must be referred, not quietly accepted.
- Whole life cost, not purchase price, is the number that should drive delivery model and award decisions.
- Governance makes models credible: documented assumptions, formal QA and separation of builder and tester.
- Build the model at planning stage, and keep it alive after award as the contract management baseline.
Want the full breakdown?
The complete explainer covers the key facts, the requirements in detail and a practical action list, free and printable in the Procurement Library.