Endorsing bodies read the financial model before they read the pitch. Envestors locks applicants into its own FCA-compliant template because it has seen too many founder-built spreadsheets that don't add up. Innovator International runs every forecast through a "would I invest?" sanity check. Both bodies reject applications where the numbers don't survive a second pass. This piece walks through eleven financial models worth building, in the order they actually matter.
Most applicants overbuild the glamorous models (DCF, Monte Carlo) and underbuild the ones the assessor reads first (three-statement, break-even, sensitivity). Invert that order.
Why financial models carry so much weight
The Innovator Founder Visa is assessed against three pillars: innovation, viability, and scalability. Two of the three — viability and scalability — live or die on the numbers. A persuasive innovation pitch with a weak financial model fails viability. A strong viability story with no scalability runway fails the £1m-in-three-years benchmark that sits behind most endorsements.
The endorsing bodies converge on a narrow set of expectations. Envestors wants a three-year forecast with P&L, balance sheet, and cash flow tied together. Innovator International wants the funding figure matched to a bottom-up cash-flow forecast with a 20% Prince2 contingency. Both want the assumptions behind every number written down in plain English next to the number itself.
It's all about conservative sales assumptions, it's really underpinning those assumptions with logic.
"Underpinning with logic" is the test every model in this article has to pass. A number without a written assumption fails. A written assumption without a cited source fails. A cited source that doesn't match the claim fails.
Model 1 — Three-statement integration
The P&L, balance sheet, and cash-flow statement linked so that a change in one line flows correctly into the other two. This is the baseline the endorsing bodies expect. An applicant who submits a P&L and no cash flow has failed viability before the first read.
How to build it
Start with a monthly P&L for three years. Tie the closing balance of each month's cash line to the opening balance of the next month on the balance sheet. Reconcile the cash-flow statement to match the change in the cash line. If the three statements don't tie, the model is broken and every downstream number is wrong.
UK accounting templates from the Financial Reporting Council and the ICAEW mirror the format Companies House will eventually see from your statutory accounts. Matching the format early is cheaper than retrofitting it at year-end.
What assessors probe
- Does operating cash flow trend to positive before runway runs out?
- Does the balance sheet show enough cash through the 24-month runway minimum?
- Does working-capital movement explain the cash burn?
Model 2 — Break-even analysis
The month where monthly revenue crosses monthly operating costs. This is the anchor number for Innovator International's funding test — Richard Harrison's position is that funding must cover costs to break-even plus contingency.
If a company needs £80,000 to get to a point where it's sustainable they have to provide that £80,000.
How to build it
Plot monthly revenue against monthly fixed costs. The crossover month is break-even. Add a 20% contingency. The cumulative burn from month one to break-even is the minimum evidenced capital you must have in the bank.
Watch-outs
Model 3 — Scenario and sensitivity analysis
One model, three runs: base, downside, and upside. Each run varies the two or three assumptions that drive the whole forecast — usually price, conversion rate, and sales-cycle length.
How to build it
Use named cells for pricing, conversion, and sales-cycle inputs. Duplicate the sheet and change only those inputs. The downside case should assume revenue starts 3–6 months later and grows at half the base pace. If the downside case still shows survival within your evidenced capital, the plan is defensible. If not, raise more capital or cut costs before submission.
Know exactly where your application stands.
Get your free AI assessment in 90 seconds.
Get your assessmentModel 4 — Cash-flow forecast (bottom-up)
A month-by-month cash-flow projection built by listing every inflow and outflow, not derived from top-line revenue assumptions. This is the model Envestors' template is built around and the one Innovator International's funding calculation runs off.
How to build it
List every cost line individually: UK salaries at market rates, London rent at £400–800 per desk per month, accountant fees at £150–300 per month, visa and endorsement fees upfront, IP filings, software, marketing budget per customer, cost of goods or service delivery per unit. Then model the revenue curve month by month — first customer in which month, tenth customer in which month — based on your sales-cycle assumption.
Why "bottom-up" is non-negotiable
Round-number capital asks (£50k, £100k, £150k) signal a top-down forecast. Assessors spot them immediately.
It's all about conservative sales assumptions, it's really underpinning those assumptions with logic.
Model 5 — Unit economics
Revenue per customer, cost to acquire that customer, cost to serve that customer, and the resulting contribution margin. The scalability pillar depends on unit economics that survive scaling.
How to build it
Three inputs: Customer Acquisition Cost (CAC), Customer Lifetime Value (LTV), and gross margin per unit. LTV/CAC above 3 is the rough benchmark for a scalable software business. Below 1 is usually unworkable. Under 2 requires explanation.
Harrison's scalability rule sits on top of unit economics:
A business should generate £100,000 per employee. So a good consultant would generate £100,000 a year. A 10-person business should generate minimum of a million a year.
See £100k per employee scalability for how this plays through to headcount planning.
Model 6 — Funding requirement and contingency
The total capital required, broken into categories, with an explicit contingency line. Straight out of the Prince2 playbook.
How to build it
Four buckets: setup costs (relocation, formation, legal, endorsement fees), operating costs to break-even, personal livability through the same period, and a 20% contingency on top of the three bucket subtotals. The total is the minimum evidenced funds you need. See the 24-month runway rule for how Envestors frames the same number.
The evidence chain
Every pound must be traceable to a bank statement, signed investment agreement with matching fund statement, or historical business revenue stream. Pledges, term sheets, and "I'll raise on arrival" do not count.
Model 7 — Revenue build (customer cohort)
Rather than one revenue number per month, model revenue as a stack of customer cohorts. Cohort one acquired in month three, cohort two in month four, each with its own retention curve and expansion revenue.
Why this matters for endorsement
A cohort model forces you to state the conversion rate, the sales-cycle length, and the churn rate explicitly. It also makes the "where does the first customer come from?" question answerable in prose, which is what the formal presentation interview will probe.
How to build it
One row per cohort, one column per month. For each cohort: size, average revenue per customer per month, retention rate, expansion multiple. Sum the rows for the monthly revenue line that flows into your P&L.
Model 8 — Headcount and UK payroll
Three-year UK headcount plan with roles, start months, and fully-loaded cost (salary plus 13.8% employer NI plus pension plus benefits plus recruitment cost amortised).
How to build it
Use UK market rates, not the founder's home-country rates. A mid-level software engineer in London costs £60k–£90k base, which is £75k–£112k fully loaded. An early sales hire in London costs £50k base plus commission. If your plan has five hires by year two at £30k each, the assessor will flag it as unrealistic.
What this model feeds
Your job-creation number for ILR settlement. Two of the six settlement achievements relate to jobs created — ten jobs, or five at higher wage levels.
Model 9 — Discounted cash flow (DCF)
Future cash flows discounted back to present value. For the visa application, DCF is mostly overkill. Include a simple DCF only if you are claiming a valuation to support an equity raise referenced in the plan. Otherwise, a three-year P&L and cash-flow is what the endorsing body wants.
How to build the simple version
Forecast free cash flow for five years. Terminal value at a 2x revenue multiple. Discount at 20–25% for early-stage risk. The output is what an investor would theoretically pay today.
Model 10 — Capital budgeting (NPV and IRR)
Only relevant if your plan involves a discrete capital decision — a factory, a lab build-out, a bulk purchase of equipment. For software and service businesses, skip this entirely.
List the cash outflow (the investment) and the cash inflows across the forecast years. NPV is discounted inflows minus outflow. IRR is the discount rate at which NPV equals zero. A positive NPV at a 15–20% discount rate clears the basic hurdle.
Model 11 — Monte Carlo simulation
A probabilistic model that runs the forecast thousands of times with randomly sampled inputs. Useful only where genuine uncertainty dominates a specific variable — drug trial timelines, hardware yield rates, commodity prices. A biotech plan legitimately uses Monte Carlo. A SaaS plan where CAC is "somewhere between £50 and £200" does not — a sensitivity table is enough.
If you can't name the input distributions the Monte Carlo is sampling from and explain why each is shaped the way it is, don't include it. Decorative Monte Carlo is worse than none — it signals innumeracy.
Which models actually matter for endorsement
Rank-ordered for a typical software or service business:
- Three-statement integration (non-negotiable)
- Break-even analysis (non-negotiable)
- Scenario and sensitivity analysis (non-negotiable)
- Cash-flow forecast, bottom-up (non-negotiable)
- Unit economics (LTV/CAC, gross margin)
- Funding requirement with contingency
- Revenue build by cohort
- Headcount and UK payroll
- DCF (only if raising equity)
- NPV/IRR (only for capital-intensive businesses)
- Monte Carlo (only when uncertainty is the dominant variable)
The first eight are table stakes. The last three are optional and often counterproductive when built decoratively.
Common failure patterns
- Top-down revenue plugged into bottom-up costs. Round-number revenue with meticulous cost lines reads as internally inconsistent.
- No downside case. The plan works if every assumption hits and fails if any one misses by 20%. See ghost-written ideas and buzzword traps.
- Home-country salary assumptions. UK wages are higher than most founders' reference markets. Underestimating them collapses the runway calculation.
- Missing contingency line. Innovator International requires it. Envestors expects it. Its absence signals inexperience.
- Decorative Monte Carlo or DCF. Models included to look sophisticated. Assessors downgrade on innumeracy.
See letters of intent vs paying customers for how revenue assumptions should be evidenced, not just modelled.
External context
The Home Office Innovator Founder Visa guidance sets the outer constraint — the business must be viable. The endorsing bodies operationalise "viable" through the financial model. Companies House filing requirements dictate the statutory accounting format your model should already resemble, so the transition from plan to live company is frictionless.
Key takeaways
- Build the first eight models before you touch the last three. Assessors read the basics first.
- Every revenue and cost number needs a written assumption next to it. A number without an assumption fails the "underpinning with logic" test.
- The break-even calculation plus a 20% contingency sets your minimum funding evidence.
- UK salary rates, UK rent, UK professional fees — use UK numbers. Home-country rates collapse the model.
- Decorative models (Monte Carlo built without distributions, DCF built without a raise) signal weak preparation. Cut them.
- financial-model
- cash-flow
- break-even
- viability
- business-plan
- templates