If all financial models are wrong, why bother in the first place?

Our Cambridge Angels member Matthew Cleevely digs into some key questions that founders ask.

“How far out should I forecast?”
“When do investors expect breakeven?”
“Over what period should revenue be projected?”

I get asked variations of these questions a lot whenever I talk about financial models. The temptation to answer “that’s Numberwang” is strong, but I’m usually more helpful than that. The more boring but much more useful answer is: if your model is mainly being built for what you imagine investors want, you are probably building the wrong kind of model. The tail is wagging the dog.

As this is finance-adjacent, I should do the usual disclaimer fun. This is not advice. You need to make up your own mind. If you do something truly stupid with a spreadsheet, that’s on you.

My more substantive disclaimer is that all financial models are wrong anyway. It’s a fact not a criticism. A model is not a crystal ball. It’s a simplified story of how cash, and sometimes value more broadly, flows in and out of your business in the future, told in numbers. You build one to answer specific questions about possible futures of your business.

The questions that matter are things like:

How am I actually going to get customers? How much will they really pay?
How many people do I need to hire, to get what done?
How much money do I need, and what exactly does it get me?
What happens if things go wrong? If I can’t raise after this round how can I manage my spend?
What things will tell me that I need to cut back?

Those are useful questions because they force you to think about what your business actually is, what drives it, what builds value, and what could kill it. A good financial model is really just a way of making those questions explicit and then forcing yourself to put a story-in-numbers next to the answers.

That is also why modelling is iterative. You build a model to answer a question for a specific  audience, look at what it tells that audience, then realise you’ve asked the wrong question or used the wrong mechanics, and go round again. My first model is usually extremely detailed, does everything and is mostly useless. Then I come back to it, realise it answered the wrong thing, throw out most of it, and build the one I actually needed. The simpler the model, the easier that iterative loop is. Your own job as founder is hard enough without building a spreadsheet cathedral you can’t explain. The circle goes questions >models>audience>answers… then back round again.

For founders, the first audience for the model is you. That matters. You need to understand the mechanics of your business and how cash moves through it. Investors come second. Employees come later, when you need targets and operating numbers. HMRC gets a special mention for EIS/SEIS and related joys. But if the founder cannot use the model to provide clarity it’s performative investor theatre - and investors don’t like that.

When I look at an early-stage model as an angel, I’m not really trying to work out whether the spreadsheet is “correct”. It won’t be. What I’m trying to work out is more basic answers to my own questions about a business:

Does this founder understand how value gets created in this business / market?
Do they understand what it costs to build, sell and deliver what they’re proposing, have they thought about it?
How much cash do they need to get to the next meaningful milestone? What happens then?

What does the revenue and value profile of this business look like and is it realistic given the plan?

Investors want to see future value expressed in numbers they can consume as a story that goes along with the pitch deck (aka a plan!). In software that value is usually easiest to express through revenue. In deep tech it may be patents, approvals, contracts, data, technical milestones, regulatory progress, or some other form of de-risking that makes later value creation believable. But the underlying question is the same: where is value coming from, how much and when and what are the activities you do to build it?

This is also why TAM, SAM and SOM can be really useful but are usually closer to bullshit decorative. They are useful if they show that you’ve anchored ambition to a real market and a credible route into it. They are decorative if they are just reaaaally big numbers in a slide deck. Most markets are huge in the abstract. What matters is if someone has understood it and their place within it and can define their specific segment and how they can build a business in it. Understood, defined, and there is a plausible mechanism for getting at it. If you understand your market, you should be able to get a realistic meaningful percentage of it (10-20%) with some reasonable sales paths. Not 0.1% of a $trn market or 100% of a $bn one.

If the honest answers are not venture-scale, you probably do not have a VC business. That is fine. It is much better to discover that in a spreadsheet than after two years of trying to contort your company into a shape it was never supposed to be. Likewise if you can find that you can grow your business without investment even better - go bootstrap!

So, back to those practical questions.

How far out should you forecast?

My answer is: in detail for as long as you need to understand when you run out of money. Usually enough to understand the next 12–24 months properly, and far enough beyond that to show what the business could become. Monthly detail matters in the short run, because that is where reality is: hiring, payment timing, sales cycles, cash burn, the point at which things start going wrong. Beyond that it’s painting a picture of the direction of travel and your understanding of it. Detailed nearer term, then longer-range visibility beyond that, but only report what you can actually narrate.

How soon do investors expect breakeven?

Usually the wrong question. Or at least the wrong first question.

Angels do not all sit there waiting for a universal breakeven date. What matters more is whether the company is using cash to get somewhere meaningful. Product-market fit. Technical de-risking. A real commercial engine. Some evidence that the thing being built will be worth materially more later than it is now. Breakeven can matter a lot in some businesses and much less in others. What always matters is whether the founder knows what the money is buying, why that creates value and to whom in the market.

Whilst you’re here, it’s worth pointing out a few common traps I see a lot in models themselves.

1) Founders get hypnotised by exact numbers. Year 5 revenue of £23,458,392.67. Lovely. Meaningless. Precision is fake confidence. Round numbers are usually better.

2) Build beautiful centreline models where everything goes perfectly. Nothing goes perfectly. Model what happens when growth is slower, CAC is worse, hiring takes longer, churn is uglier, or the next round arrives six months late. That is sensitivity analysis: what does the business look like when things go wrong? It is one of the few ways you can tell whether you understand what actually breaks it.

3) Assume revenue scales while costs politely stay where they are. They do not. Costs scale often much faster than people expect - go look at scaled industry players and their cost bases, it’ll give you a good idea of a realistic mix of your business ‘at scale’.

4) Complex, brilliant, and completely useless model: incredibly detailed, beautifully formatted, and completely unusable. If you can’t explain it, simplify it until you can.

So what does “good” look like?

Honestly, much simpler than most founders think. A summary tab. The key question stated. Assumptions visible. Outputs visible quickly. At least one reality check. Something you can actually change and see the result from. Someone else can understand it. And you can describe what it shows in a few sentences. That’s a decent model. It can be as little as a single table with entirely manual numbers. It does not need to be a spreadsheet opera - lots of founders think it needs to be, it’ll get there, but that’s when you can afford a CFO and some accounting agents.

So all financial models are wrong. But that doesn’t make them useless. They are tools for asking better questions, understanding where reality bites, and deciding what kind of business you are actually trying to build and how to resource it.