Guide · 7 min read · Updated 2026-07-25
How Much Contingency Should a Construction Project Carry?
Ask ten estimators what contingency to carry and most will say five per cent, because that is what the last job carried, and the one before it. But contingency is not a custom. It is a price for the risk that remains after you have priced everything you can see, and that risk is wildly different between a repeat fit-out with finished drawings and a refurbishment behind unopened walls.
This guide covers what actually drives contingency, sensible starting ranges, and why the industry systematically carries too little.
What contingency is, and what it is not
Contingency is money for known-unknowns: risks you can name but not fully price, like ground conditions, design development and market movement. It is not a substitute for pricing the scope properly, and it is not the same as profit, though on badly run jobs it quietly becomes the first place profit goes to die.
Keep three pots distinct in your head: estimating allowances inside rates for normal waste and variance, contingency for named risks, and margin. When the three blur, the job looks cheap at tender and surprises everyone at final account.
The six drivers that should set the number
- Design maturity: pricing concept drawings deserves multiples of the contingency of pricing construction-issue drawings
- Ground and existing-structure risk: anything below ground or behind existing finishes carries the largest single unknowns
- Contract terms: a fluctuations clause, or the lack of one, decides who owns inflation; onerous amendments shift risk to you and belong in the number
- Procurement route: single-stage lump sum on immature design piles risk on the contractor; two-stage or cost-plus shares it
- Programme and season: compressed programmes buy risk, and weather-exposed work carries more of it in winter
- Counterparty strength: a weak payer is a risk to the whole revenue line, not just the margin
Starting ranges, honestly framed
There is no universal correct percentage, and anyone quoting one without seeing the job is guessing. But as commonly used starting points, before adjusting for the drivers above:
| Situation | Common starting range |
|---|---|
| New build, mature design, known ground | 3 to 5% |
| New build, developing design | 5 to 10% |
| Refurbishment or work to existing structures | 10 to 20% |
| Concept-stage pricing or heavy below-ground risk | 15% and up |
Treat these as the start of the conversation, not the end. The right number for your job comes from the drivers, and from your own outcome history on similar work.
Why the industry carries too little
Optimism is structural. Tenders are won by the lowest credible number, so every pressure at bid stage pushes contingency down. Research by RICS and CIOB has found that most UK projects that overran had insufficient contingency at tender, and Bent Flyvbjerg’s global project database shows cost overrun is the norm on large projects, not the exception. Your own last three jobs are probably better evidence than either.
The correction is to size contingency from evidence rather than mood: what did similar jobs actually cost against their tender, and what does that history say the tail risk looks like?
A better way than a flat percentage
A single percentage hides the real question: how confident are you? Probabilistic forecasting answers it directly. A Monte Carlo simulation runs the job thousands of times with the risks allowed to vary, and reads off the cost at chosen confidence levels. The P80 figure, the cost you come in under 80 times in 100, is a far more honest basis for contingency than a flat percentage, because it is derived from the job’s own risk profile.
This is what BuildPredict does with the documents you already have: it reads the tender, prices the risk signals it finds, and returns a P50 to P90 cost range with a suggested contingency in pounds. You can try it on your next tender with the free pre-bid check.
Frequently asked questions
What is the difference between contingency and a risk allowance?
Usage varies, but a common split: risk allowances price specific named risks individually, while contingency covers the remaining unallocated uncertainty. What matters is that the pots are explicit and nobody double-counts or quietly deletes them at adjudication of the bid.
Who owns the contingency, client or contractor?
Both usually hold their own. A client carries a project contingency for scope and employer risks; a contractor carries pricing contingency for delivery risk they own under the contract. Problems start when each side assumes the other is holding the money for the same risk.
Should contingency be visible in the bid?
For lump-sum competitive bids it is normally embedded in rates, because a visible line invites deletion. In two-stage or open-book routes a transparent risk register with priced items usually serves you better and builds trust. Match the transparency to the procurement route.
What is a P80 cost forecast?
It is the cost you would expect to come in at or under in 80 out of 100 runs of the job, taken from a probabilistic model such as a Monte Carlo simulation. Budgeting at P80 rather than the average makes an overrun the exception rather than the coin flip.
This guide is general information for UK construction businesses, not legal advice. Contract terms differ, and for significant disputes or heavily amended contracts you should take professional advice.