Revenue forecasting for startups without a data team

A trend line drawn through total revenue hides its parts: a strong sales month can cover a rise in churn until the churn is too large to cover. Forecasting each part on its own shows which one moved.

Each part gets a low, a base and a high value, taken from your slowest, typical and best months. The forecast is the range they produce, and it says which input moves the result most, so you know which number to pin down first.

With less than a year of history, the inputs are mostly assumptions, and the forecast names them. Each month, last month’s forecast is set against what happened, part by part, and the inputs are updated.

• Monthly recurring revenue, split into new, expansion, downgrades and churned.
• Customer churn and revenue churn, each month: the share of customers who left, and the share of revenue lost to cancellations and downgrades.
• Net revenue retention: what last year’s customers pay now, against what they paid then.
• Gross margin.
• What a customer costs to win, and the months of margin it takes to earn that back.
• Cash burn and runway.

Each KPI needs a written definition and enough records behind it. A monthly figure built on a handful of records moves with whichever records arrive, as one of our own reports showed, and a dashboard can count something other than what its label says, as our own web analytics did.

Benchmarks are left out on purpose: what good looks like depends on your market and your prices, so the first comparison is with your own months.

Each one is defined exactly, with a worked example and the place it usually goes wrong, in What KPIs should a Series A startup track?

Not covered. Building a data warehouse, or dashboards in Looker or Power BI. The report reads the data where it already is, and nothing is installed on your systems.

An example. A monthly revenue report for an invented subscription company shows what arrives each month: the figures, the revenue bridge, the forecast and the checks that ran. It is one of the examples of what a client receives.

Who it is for. Companies with no data team of their own, with revenue in a billing system or a warehouse, who need a forecast for the board and a report that arrives each month.

The questions it answers. What revenue will be in six and twelve months, and how wide the range is. Which input moves it most. Which KPIs to track, and how each is defined. Why last month missed the forecast.

What you need to have. Billing or invoicing history per customer, with plan changes and cancellations; a list of new customers by month; costs, at least by month; and your current plan or budget, if one exists.

The method, in plain words. Rebuild monthly recurring revenue from the billing records and check it against what you invoiced. Split it into its parts. Measure each part’s range from your own months. Run the forecast, find the input that moves it most, and write the KPI definitions down.

What you receive. A forecast memo of up to two pages, with the range and the inputs behind it. The KPI definitions. A recurring report, as a document, a notebook or a simple web page. A method note listing the checks that ran.

How it is checked. Agents build the forecast and the report. Every deliverable gets number and claim checks run by scripts, not AI, and a review by an agent. On what you act on, such as a model going live or a decision memo, that review is a panel of three reviewer agents and a judge, and then the analytics director signs it; routine refreshes do not need a sign-off. The checks are described on how it works.

Your data. Where it goes, which model providers process it and what happens when the work ends: see data handling.

Price. Fixed per deliverable. A forecast with its KPI definitions, or the first build of a recurring report, is a sprint, from €12k, fixed in a one-page proposal before work starts. Repeat work runs on an async retainer, from €8k a month.

Who builds it. FinDaS AI’s analytics director built reporting functions from scratch, and worked on budgeting and forecasting, during eight years in retail-banking and consumer-finance CRM analytics. Forecasting is one of the money questions the same team works on.

A subscription software company, just after its Series A, starts the year at €308,000 of monthly recurring revenue from 4,000 customers, €77 each on average. New customers join on its €29 Starter plan, as do customers who come back. Its last twelve months give each input a low, base and high value.

The base case grows recurring revenue by 4% in a year; the range runs from a fall of 8% to growth of 17%. A single line at €320k would hide that month-12 revenue could land anywhere from €282k to €361k.

Moved one at a time, with the others at base, the range in new and returning customers moves month-12 revenue between €312k and €328k, leaving between €303k and €338k, and what existing customers add between €306k and €334k. So the first number to pin down is churn; the churn prediction page covers it. New customers move revenue least here, because they join on the €29 plan.