Two forecasts, no black box: MRR and cashflow, built from your own data

Most subscription-analytics tools treat forecasting one of two ways. Either they don't do it at all, or they hand you a number out of a black box — a projection you can't inspect, can't reproduce, and therefore can't really defend in a board meeting.

We just shipped two forecasts, and neither is a black box. Both are built entirely from your own movements, and both are transparent enough that you could reproduce them in a spreadsheet if you wanted to. They answer two different questions, and I think you need both.

The MRR forecast: where you're headed, and what would change it

The MRR forecast projects your recurring revenue forward and lets you stress-test it with three levers: new customers, net revenue retention, and price. Drag a lever and the scenario line moves away from your baseline in real time — no spreadsheet, no assumptions you can't see.

Under the hood it's a plain customer-count model, the same shape used across the industry:

Next MRR = this MRR × monthly NRR + (new customers × average sale price)

Every input is the average of your own recent months — new customers per month, ASP, monthly retention — and each one is shown right under its slider, so you can always trace a projection back to the actuals behind it. If you're growing fast, a six-month average will feel conservative, so you can set the baseline window to the last 3 months and it tracks your current momentum instead of a slower past.

I want to be honest about what this is: a trailing-average model, not a crystal ball. It won't predict a fundraise or a viral month. The levers are how you steer it toward what you actually believe. What it will do is give you a number you understand — which, for hiring and board decisions, beats a fancier number you have to take on faith.

The cashflow forecast: when the money actually lands

MRR tells you how much recurring revenue you've earned. It says nothing about when the cash hits your account — and those are very different things.

The cashflow forecast answers the second question. It walks every active subscription forward along its own billing schedule and drops each renewal into the period it actually bills. So an annual plan lands as one payment on its renewal date — a real spike — instead of being smoothed across twelve months the way MRR smooths it. That lumpiness is the whole point: it's the shape your bank account actually sees.

It's a committed-cash view. It projects the book you have today — it doesn't invent new customers, because you can't collect cash from customers you haven't signed. And because real customers pay late and some cards fail, there's an optional toggle that shifts each customer's cash by the delay we've measured on their invoices and deducts the share of payments that fail and are never recovered. Turn it on for a conservative, planning-ready number.

This is the view you want for runway, hiring, and spend decisions — the questions where "how much MRR do we have" is the wrong denominator.

Why two forecasts, not one

Because they answer different questions:

• The MRR forecast is what could this business become? — it models growth, retention and pricing, and lets you play out scenarios.
• The cashflow forecast is what will my current book actually pay me, and when? — committed cash, on the real billing calendar.

One is about ambition; the other is about the bank. You need both, and blending them into a single number would ruin both.

Forecast any slice, not just the whole company

Here's a detail I'm quietly proud of: both forecasts respect your filters. Narrow the report to a single plan, a country or region, a saved segment, or any custom variable, and the entire forecast recomputes for just that slice — baselines, levers, projected cash, the lot. Want to know the cash your annual plans will bring in next quarter, or how MRR would grow if you only counted enterprise customers in the US? Filter, and the forecast follows. It's the same filtering you already use on every other report, so there's nothing new to learn.

The part I care about most: you can check our work

We deliberately chose transparency over sophistication. There are more elaborate models we could have shipped — ones that would forecast marginally "better" on a backtest and that you'd have to trust blindly. For the decisions these forecasts actually drive, I'll take the number you can defend every time. Every figure ties back to your own MRR and movements. If a projection looks wrong, you can find out exactly why.

Both forecasts are available on every plan, including the free one — same as everything else we build. Connect Stripe, Chargebee or Recurly and you'll see them in minutes.

Try the forecasts on your own data, or take a look at the live demo. As always, I'd love to hear what you think.