A fixed fee is a promise: whatever this job turns out to involve, you'll deliver it for the number on the proposal. Your client's cost is capped. Yours isn't.
This post is for MDs and partners of small technical consultancies — structural, civil, M&E, geotech, environmental — who bid most of their work at fixed fees. It covers how to manage bid risks properly: how to see the risk you're taking on before you price it, how to load it into your fee, and how to hold the line when a client asks you to sharpen the pencil.
Because the data says this is where firms get hurt.
What does the insolvency data actually say?
The Building Cost Information Service (BCIS) published its latest insolvency analysis on 2 June 2026. In March 2026 alone, 347 construction firms became insolvent — 16% of all insolvencies in England and Wales. Over the twelve months to March, the total was 3,827. That's 7% down on the previous year, but still 19% above pre-pandemic 2019.
347
Insolvencies, March 2026
16% of all insolvencies in England and Wales
3,827
Twelve months to March
7% down on the previous year
+19%
vs pre-pandemic 2019
Still well above the old baseline
The line worth reading twice is from BCIS chief economist Dr David Crosthwaite:
Persistent cost pressures, tight margins and cash flow challenges still affect businesses across the supply chain.
— Dr David Crosthwaite, BCIS chief economist
He singled out specialist firms as particularly vulnerable — because of fixed-price contracts and payment delays.
That's a neutral authority, looking at the whole market, naming the mechanism. Not "bad firms fail." Not "the market is tough." Specifically: firms that commit to a fixed price, then absorb whatever the job actually costs, are the ones that run out of road.
Note that Crosthwaite named two mechanisms, not one: fixed prices and payment delays. They compound. A mispriced job erodes margin; slow payment turns that eroded margin into a cash-flow problem, because your engineers' salaries go out monthly whether or not the developer's certificate has landed. A thin fee on 60-day terms is a very different proposition from the same fee paid monthly — which is why payment terms belong in the bid decision, not just the contract review.
To be clear about what this means for you: the data describes the market, not your firm. Most consultancies that misprice a fixed-fee job don't go under. They do something quieter — they work the overrun at nights and weekends, burn the margin, and call the job a win because the invoice got paid.
Why do fixed-price bids go wrong so often?
Because the risk in a fixed fee is asymmetric, and most pricing exercises ignore that.
When you estimate a job, you're estimating a range. Fifty-two delivery days is really "somewhere between 46 and 70, depending on how the design develops." But a fee is a single number. So somewhere in the pricing process, the range collapses to a point — and it almost always collapses to the optimistic end, because the optimistic end wins work.
Now look at who holds which side of the bet:
▼Job comes in under estimate
- The client doesn't pay you more.
- You keep a slightly better margin.
▲Job comes in over estimate
- The client doesn't pay you more either.
- You eat every extra day.
Either way, the client's cost is capped. Only yours moves.
You've sold the client a cap on their cost. In financial terms, that cap is worth something — it has a price. Most small consultancies hand it over for free, then wonder why "won" jobs keep coming in thin.
The overruns themselves are rarely exotic. For a typical consultancy package they're the same four or five suspects every time: revision cycles beyond the number you assumed, coordination drag with other designers, scope that was vague in the brief and generous in the client's memory, "reasonable attendance" clauses that turn into standing weekly commitments, and information arriving late so work gets done twice.
None of those are surprises. They're just unpriced.
How do you price risk into a bid?
You don't need a quantitative risk model or a Monte Carlo simulation. You need fifteen minutes, honesty, and three columns.
Before the fee goes on the proposal, list the specific ways this job could take longer than the base estimate. Not generic risks — this job's risks. Then, for each one, write down two numbers: roughly what it would cost you in extra days if it happens, and roughly how likely it feels. Multiply. Add the results up. That total is your risk loading.
Is this crude? Yes. A 50% probability estimated over coffee is not science. But crude and written down beats precise and imaginary — and it forces the conversation a fee should force: which of these risks are we carrying, and which are we excluding?
That last question is where the real money is. Every risk you identify has three possible homes:
Priced in
You carry it, and the fee reflects it.
The right home for risks too small or too likely to argue about.
Excluded
The proposal says plainly that it's not included, with a day rate for when it arises.
Revision cycles beyond two is the classic.
Capped
You carry it up to a stated limit.
"Attendance at up to four design team meetings" is a cap. "Reasonable attendance" is a blank cheque.
A fixed fee with clean exclusions is still an easy yes for the client. It's just no longer an uncapped bet for you. (The usual hedge applies: how risk is allocated in your appointment is a contract matter — this is operational guidance on pricing, not legal advice. If a clause worries you, that's a conversation with your professional indemnity insurer or a lawyer, not a blog post.)
What does a risk-loaded margin floor look like?
Here's the arithmetic on a job sized for a firm like yours.
A structural package for a residential developer. Fixed fee on the table: £45,000.
Base cost, the version most spreadsheets stop at:
| Input | Figure |
|---|---|
| Delivery days | 52 |
| Blended day cost | £620 |
| Labour cost | £32,240 |
| Non-recoverable expenses | £1,400 |
| Base cost | £33,640 |
On that maths your floor — the fee below which the job loses money — is £33,640. At £45,000 you're carrying £11,360 of margin, about 25%. Comfortable.
Now the fifteen-minute risk pass:
| Risk | Extra cost if it hits | Likelihood | Expected cost |
|---|---|---|---|
| Revision cycles beyond the two assumed | 6 days · £3,720 | 50% | £1,860 |
| Coordination drag with M&E redesign | 4 days · £2,480 | 40% | £992 |
| "Reasonable attendance" during construction | 5 days · £3,100 | 60% | £1,860 |
| Risk loading | ≈ £4,700 |
Your risk-loaded floor is £38,340. Still £6,660 of genuine headroom at the quoted fee — a real margin, honestly stated at roughly 15% rather than flattered at 25%.
Now the phone rings. The developer likes you, but a competitor came in lower. Could you do it for £40,500 — a 10% trim?
Against the naive floor, £40,500 looks fine: nearly £7,000 clear. Say yes, and if those risks land at expected value your actual margin is £40,500 − £33,640 − £4,700 = £2,160. About 5%. One bad week from zero. And if all three risks hit in full — £9,300 of overrun — you deliver the job at a £2,440 loss, while your bookkeeping insists you won a profitable project.
Against the risk-loaded floor, the same request reads completely differently: the discount doesn't trim your margin, it consumes almost all of it. Same job, same client, same £4,500 — but now the decision is visible before you make it.
The £45,000 bid, on one bar
£33,640
Naive floor
£38,340
Risk-loaded floor
£40,500
Discount ask
£45,000
Quoted fee
Run the risk pass before you first write the fee down, not after. Once a number has been said out loud — internally or to the client — every subsequent calculation quietly bends toward justifying it.
How does a floor change the conversation with the client?
The floor's first job isn't in the negotiation. It's ten minutes earlier, in your own head — replacing "can we win at this number?" with "is this number worth winning at?" Those are different questions, and under fee pressure the first one always shouts loudest.
Once you know the floor, discount requests stop being tests of nerve and become arithmetic:
- Above the floor, you can trade. And you can trade properly — a lower fee in exchange for a tighter scope, an extra exclusion, or better payment terms. You know exactly what the concession costs, so you can ask for something real in return.
- At the floor, you can say yes with your eyes open — as a deliberate, one-off decision about a client relationship, not a drift.
- Below the floor, the answer is a calm no. Not "let me see what I can do." A specific number you can stand behind: "Below £38,500 this job doesn't work for us at the level of service you need."
That last sentence does more for your credibility than any discount ever will. Engineers respect a firm that knows its own numbers. And if the work goes to whoever priced none of these risks — the BCIS insolvency tables suggest how that strategy tends to end. You don't need to win the jobs that only exist below your floor.
The prerequisite is that the floor is decided before the pressure arrives. A walk-away number invented mid-negotiation isn't a floor, it's a mood.
What should you do before your next bid?
The discipline is small enough to start this week:
- Estimate the base cost honestly — days × blended day cost, plus expenses.
- Spend fifteen minutes on the risk list — this job's specific overrun paths, costed and weighted.
- Decide each risk's home — priced, excluded, or capped, and make the proposal say so.
- Write the floor down — base cost plus risk loading, agreed before the fee is quoted, visible to everyone who might get the "sharpen your pencil" call.
- Log the outcome — when the job closes, compare actual days against the estimate. Your risk percentages stop being guesses after a dozen jobs. That same discipline feeds a better bid/no-bid decision on the next enquiry, because you'll know which clients and job types keep blowing through their estimates.
If your bids currently live in a spreadsheet where no floor column exists, that's the gap Bidro is built to close: every bid gets a floor before it gets a fee. We've written up where a bid tracking spreadsheet stops working — including the cases where it's still the right tool.
The firms in the insolvency data didn't fail because they couldn't do the engineering. Plenty failed because they kept promising fixed prices for variable work and hoping the difference would be kind. Price the risk. Know your floor. Then bid like you mean it.
Sources: BCIS insolvency analysis, 2 June 2026; BCIS construction insolvencies tracker. Worked figures are illustrative examples, not market benchmarks.