Every investor knows your three-year projections are wrong. You know it too. The number in year three of a seed-stage model has never once matched what actually happened.

So why does everyone still want to see them? Because they’re reading the model for something other than the number: which variables you think drive the business, whether you understand what your own growth costs, and whether the story in your deck and the numbers in your model match.

Here’s what credible three-year financial projections contain, the assumptions investors test first, and the mistakes that cost founders credibility in the room.

What financial projections are actually for

Your financial projections do three jobs in a raise. They size the opportunity, they justify the amount you’re asking for, and they demonstrate financial judgement.

The third is the one founders underinvest in. An investor comparing two startups with similar traction will back the founder whose model shows they know what each new customer costs to win and serve. They’re assessing the founder as much as the model.

The structure a credible model has

Three linked statements – profit and loss, balance sheet and cash flow – driven from one set of assumptions on a separate tab. We’ve covered the mechanics of that in How to Build a 3-Statement Model That Drives Real Decisions.

What matters for a raise is that the assumptions are visible and changeable. If an investor asks what happens when churn is 3% instead of 1.5%, you should be able to change one cell and show them, live. A model where that requires rebuilding anything tells them it was made for the pitch rather than for running the business.

Alongside that, three cases: base, downside and upside. These should be three sets of decisions rather than three revenue numbers, so the downside case needs to show what you’d stop spending on and when. That’s what an investor is really asking about.

Build revenue bottom-up, sense-check it top-down

Bottom-up means building revenue from the units you control: leads, conversion rate, customers, average contract value, churn. It produces a number you can defend line by line and a plan you can actually execute against.

Top-down starts from market size and takes a share of it. On its own it reads as wishful, and the “we only need 1% of a £4 billion market” line is the fastest way to lose a room.

Use both. Build bottom-up, then check the result against the market – if your bottom-up model has you at 18% share of the UK market by year three, one of your conversion assumptions is wrong.

The assumptions investors test first

Customer acquisition cost, fully loaded. Not just ad spend. Sales salaries, commission, marketing tooling. Models that show CAC falling steadily as you scale get challenged immediately, because it usually rises as you move past the easiest customers.

Churn. Small changes compound a long way over three years, and early-stage churn figures are usually based on too few customers to mean much. Show the number and say honestly what it’s based on.

Sales cycle length. If you’re selling to enterprise or into the public sector, the gap between signing and cash is a real constraint on how fast the money can be deployed.

Headcount and the cost on top of it. Employer National Insurance, pension and recruitment fees add meaningfully to every salary line. A plan that budgets gross salary alone understates the cost of the team you’re describing.

Gross margin. Hosting, support, payment processing and third-party licences all scale with revenue. A model where margin improves every year needs a reason attached to it.

How much detail your financial projections need each year

Year one monthly, years two and three quarterly or annually.

Monthly detail in year three works against you, because it implies a level of precision that nobody actually has that far out. What you want in the later years is a clear driver structure: if the customer number grows like this, here’s what revenue, cost and headcount do.

Tie year one to your actual cash position, so the model and your runway agree. Investors do check that the raise amount and the plan match, and a mismatch is a hard question to answer on the spot.

The mistakes that cost credibility

Hardcoded numbers inside formulas, so nothing can be tested. Revenue that grows at a fixed monthly percentage with no driver behind it. Costs that stay flat while revenue triples. A hiring plan that doesn’t appear in the cash flow. Round numbers everywhere. And the common one – a model that no longer matches the deck, because the deck was updated and the model wasn’t.

We’ve written separately about the red flags investors look for in a financial model, and most of them come back to the same thing: the model should look like something you use to run the business, rather than something built for the pitch.

The takeaway

Build the model you’d use to run the business, then show that one. Drivers visible, downside case included, year one monthly and tied to your real cash position. The number in year three matters far less than whether an investor believes you understand what it depends on.

Building the model behind your raise?

Disclaimer: This article is general in nature and does not constitute financial advice. What investors expect varies by stage and sector – speak with a qualified adviser (hey, that’s us!) about what applies to your situation.