Revenue is the hardest line in any startup forecast and the one everything else leans on. Get it wrong and your hiring plan, your runway and the size of your raise are all wrong with it.

The usual approach is to take last month and add a percentage. It’s quick, it produces a reassuringly smooth curve, and it survives roughly one investor question – because the moment someone asks what’s driving the growth, there’s no answer underneath it.

In this blog, we’ll explain why the percentage method falls apart, how to build a revenue forecast from the drivers you actually control, and how to test it before an investor does it for you.

Why a growth percentage isn’t a revenue forecast

A percentage describes what happened. It doesn’t explain why, which means it can’t tell you what to do differently.

When an investor asks what drives the 12% monthly growth in your model, “it’s what we’ve been doing” is not an answer. When you miss it, you have no way of knowing which part broke – fewer leads, worse conversion, smaller deals, faster churn – because the model never separated them.

The deeper problem is that it can’t be acted on. A driver-based forecast tells you that you need 140 more qualified leads next quarter. A percentage tells you to grow 12% and leaves you to work out how.

Building revenue from drivers

Work backwards from the revenue to the activity that creates it. For most B2B startups the chain is: leads, qualification rate, conversion rate, average contract value, time to close – then retention and expansion on top of the base you already have.

Each of those is something you either measure or can estimate from a small sample, and each one has an owner. That’s what turns the forecast into something your team can actually work towards.

For recurring revenue, build it in three parts and never merge them: opening recurring revenue, plus new, less churn, plus expansion. Presenting only the net movement hides the case where strong new business is masking a retention problem – the situation described in Subscription Revenue at Scale.

For one-off or project revenue, forecast it deal by deal with a probability and an expected close date. Weighted pipeline is fine for planning. It is not fine for cash flow, where you should only count what’s contracted.

What changes as you mature

Pre-revenue and first customers. Forecast named deals only. There is no rate to apply to a base of three. A named-deal forecast that turns out wrong still tells you more than a smooth curve that was never based on anything.

Repeatable motion, small numbers. Move to a driver model, but keep the sample size in mind – a conversion rate from twenty opportunities is indicative, not reliable. State the sample alongside the number.

Scaling. Add cohort behaviour. Customers acquired through different channels retain and expand differently, and once you have enough of them the differences are large enough to model separately.

Testing your own revenue forecast

Four checks before it goes anywhere.

The capacity check. Does the plan require more selling than your team can physically do? If the model needs each salesperson closing eleven deals a month and they currently close four, the plan depends on hires you haven’t budgeted for.

The market check. What share of your addressable market does year three imply? If it’s implausible, a conversion assumption is wrong.

The backtest. Run your driver model against the last two quarters. If it doesn’t reproduce what actually happened, it won’t predict what’s next.

The downside. Halve new business and double churn. If the business still functions, you have room. If it doesn’t, you’ve found the thing to watch weekly.

These are the same tests an investor applies, which is why doing them first changes how a projections conversation goes.

A worked example

A startup forecasts £1.2 million of recurring revenue by year end, growing 10% a month from £340,000.

Built from drivers instead: 180 leads a month, 30% qualified, 18% of those converting, at an average £9,000 annual contract value. That’s roughly ten new customers a month, or £7,500 of new monthly recurring revenue – and against an opening base churning at 1.8%, the year-end figure comes out at £790,000, not £1.2 million.

Neither number is a fact. But the second one names exactly what would have to change to reach £1.2 million: roughly 90 more leads a month, or a conversion rate above 27%, or contract values a third higher. That’s a plan the sales team can be given.

The takeaway

Build revenue from the activity that creates it, split recurring into new, churn and expansion, forecast one-off deals individually, and test the result against your own capacity before an investor does it for you.

Want a forecast that survives the questions?

Disclaimer: This article is general in nature and does not constitute financial advice. Forecasting approaches vary by business model and sales motion – speak with a qualified adviser (hey, that’s us!) about what applies to your situation.