Scenario Planning and Stress Testing for UK SMEs
Three columns labelled best, base and worst are not a scenario model. A useful one changes a small number of named drivers, and tells the board at what point it must act.
Scenario planning fails in SMEs for a boring reason: the scenarios are built by nudging the output rather than the input. Someone takes the base forecast, multiplies revenue by 0.9, and calls it a downside. Costs do not respond, headcount does not respond, and the resulting case is arithmetically impossible — which the board senses even if it cannot name why.
Start from drivers, not totals
A driver is something the business does or the market does to it, from which revenue and cost follow. Pick six to ten and no more.
| Business type | Primary drivers | Cost response |
|---|---|---|
| Professional services | Billable heads, utilisation %, average day rate | Salary base, contractor spend, recruitment |
| Subscription software | New logos, ARPU, gross churn, expansion | Hosting, support headcount, sales commission |
| Distribution / product | Units, average selling price, input cost, stock turns | COGS, freight, warehousing, buffer stock |
| Field services | Jobs per engineer, first-time-fix rate, travel time | Engineer headcount, fleet, subcontract |
Build the downside so the costs move too
- Flex the driver, not the revenue line — reduce utilisation or units, and let revenue fall out of it.
- Split costs into fixed, step-fixed and variable, and make each behave accordingly.
- Model the lag honestly: headcount reductions take one to three months and carry a cost.
- Push debtor days out in the downside — customers under pressure pay slower, and this compounds the cash effect.
- Recalculate the cash position weekly, not monthly, in the stressed case. Monthly averages hide the trough.
Macro assumptions worth stating explicitly
- Interest rate path — hold the current Bank of England base rate flat in the base case rather than assuming cuts you cannot control.
- Wage inflation, which for UK SMEs has consistently run ahead of headline CPI and drives the largest cost block.
- Employer's National Insurance and pension auto-enrolment costs on any modelled headcount change.
- Energy and rent step-ups at lease or contract renewal dates rather than smoothed across the year.
State the source and the date for each. An assumption without a provenance is an opinion, and it is the first thing a diligence process will challenge.
Covenant headroom and trigger points
For any business with debt, the scenario output that matters is not profit but headroom against each covenant, shown by test date. Present it as a table with the covenant, the threshold, the projected value in each case, and the percentage headroom. Where the downside breaches, say so plainly and bring the mitigation to the same meeting.
Then define triggers in advance: the observable event at which a mitigation is executed rather than discussed. "If two consecutive months land more than 8% below base, we pause the Q4 hiring plan" is a decision the board can pre-approve. Made in advance, it is governance; made in the moment, it is panic.
Back-test before you present
Run the model on the last four quarters of actuals and show the variance. A model that would have predicted the last year within a few points earns the board's trust in its projection of the next one. One that would not should be fixed before it is used to make decisions.
Doing this in MouCFO
MouCFO's scenario module takes unlimited named assumptions, applies them to drivers rather than outputs, and propagates the result through the P&L, balance sheet, 13-week liquidity view and covenant tracker — with each case exportable into the board pack alongside the commentary explaining what changed and why.
Frequently asked questions
How many scenarios should an SME model?
Three for the board — base, downside and a severe but plausible stress — plus any deal-specific case. More than four and the discussion becomes about the model rather than the decision.
What is a reasonable downside to model in 2026?
For most UK SMEs: revenue 10 to 15% below base, gross margin down one to two points, debtor days out by ten, and financing costs held at the prevailing base rate rather than an assumed cut. The severe case doubles the revenue shock and adds a single large customer loss.
What makes a lender trust a forecast?
Explicit drivers, a documented link between assumption and output, back-testing against the last four quarters of actuals, and a covenant headroom calculation under each case. Precision matters less than traceability.