What a subscription forecast has to get right
A subscription book is not a revenue line that grows by a percentage. It is a population of contracts, some of which leave every month, some of which get bigger, and to which you add new members at whatever rate your sales and marketing produce. Forecasting it by applying a single growth rate to last month's number hides all three of those forces and will be wrong in a specific, predictable way: it overstates the near term and misses the ceiling entirely.
The recursion this calculator uses treats each force separately. Start with the opening MRR. Remove gross revenue churn. Add expansion. Then add the new business signed in the month. Repeat. Two properties of that structure matter more than any input value.
First, churn is proportional and new business is not. Churn takes a percentage of a growing base, so its dollar cost grows with you, while a flat monthly acquisition rate adds the same dollars every month. That is why a business with steady adds and non-zero churn approaches a ceiling rather than growing forever: the ceiling is roughly new MRR per month divided by the net churn rate. At $5,000 of new MRR a month against 2% net churn, the book converges toward $250,000 — and no amount of patience gets it past that without changing an input.
Second, the retention factor (1 − churn + expansion) decides whether you are pushing uphill. Below 1.00 the existing book shrinks and new business has to cover the loss before it can add anything. Above 1.00 the book grows without a single new customer. That single number is the most important line in any subscription model, which is why it is the first step shown in the calculation.
The formula, term by term
Each month applies four operations in a fixed order, and the order is part of the model.
Churn removes c × opening MRR. Use gross revenue churn here — cancellations plus downgrades, expressed as a share of the opening book. If your only figure is a customer churn rate, it is not the same thing: losing 3% of accounts costs you 3% of revenue only when the accounts that leave are exactly average, and in practice small accounts churn more often than large ones. Measure it with the churn rate calculator and enter the revenue version.
Expansion adds e × opening MRR from upgrades, seat additions and contractual price increases inside accounts you already have. Together with churn it gives the monthly net revenue retention factor, and raising that factor to the twelfth power gives the annual net revenue retention figure investors ask for.
New business adds A × a: the accounts you sign this month times their average revenue per account. Keeping the count and the price separate is worth the extra input, because they are controlled by different teams and they fail differently — a fall in new logos and a fall in new-customer ARPA both slow growth, and the fix for each is different.
Acquisition growth raises A by a compounding percentage each month, which is what actually happens when a marketing budget or a sales team is scaling. Set it to zero to see the ceiling effect described above in its pure form.
Finally, the compound monthly growth rate summarises the whole path in one number: the constant monthly rate that would take you from the opening MRR to the ending MRR over the same horizon. It is the honest way to describe growth over several months, because it is immune to a single spectacular month.
Worked example: $20,000 of MRR, 10 new customers a month, 2% churn
Take a book of $20,000 MRR. You sign 10 new customers a month at $100 each, gross revenue churn is 2%, expansion is 1%, and acquisition is flat. Work the first three months by hand.
- Retention factor. 1 − 0.02 + 0.01 = 0.99. The existing book shrinks by 1% a month on its own.
- Month 1. Churn is $20,000 × 0.02 = $400. Expansion is $20,000 × 0.01 = $200. New MRR is 10 × $100 = $1,000. Net new is $1,000 + $200 − $400 = $800. Ending MRR = $20,800.
- Month 2. Churn $20,800 × 0.02 = $416. Expansion $208. New $1,000. Net new $792. Ending MRR = $21,592.
- Month 3. Churn $431.84. Expansion $215.92. New $1,000. Net new $784.08. Ending MRR = $22,376.08.
Notice the pattern: net new MRR falls every month — $800, then $792, then $784.08 — while nothing about your sales performance changed. The dollar cost of 1% net churn grows as the book grows, and it eats an increasing share of a constant $1,000 of new business.
Push it out and the arithmetic gives you the ceiling directly. New MRR of $1,000 a month divided by a 1% net churn rate is $100,000: at that level, the 1% monthly loss is exactly $1,000 and the book stops moving. Getting to $200,000 on this cost structure is not a matter of time — it requires either more new business each month or a better retention factor.
Now change one input. Raise expansion from 1% to 2%, so the retention factor is exactly 1.00. The existing book stops leaking, every dollar of new business sticks, and the projection becomes $20,000 plus $1,000 a month with no ceiling at all. That is the entire argument for investing in expansion revenue, in one line of arithmetic.
How to read the forecast
Check three things before you believe any subscription projection, including this one.
Is the retention factor above or below 1.00? Everything else is secondary. A factor of 0.99 caps the business at new MRR ÷ 0.01; a factor of 1.01 means the book compounds by itself. The annualised version of the same figure is what a growth investor will ask for, and 100% is the line that separates a business that grows from its existing customers from one that must keep buying growth.
Is the CMGR one you have actually achieved? Compare the calculator's compound monthly growth rate against your realised rate over the last six months. If the forecast implies 6% a month and you have delivered 2%, the model is a target, not a projection. Both are legitimate documents — but a board should be told which one it is looking at.
Where does the new business come from? A forecast that leans on acquisition growth is implicitly a spending forecast. Multiply the incremental customers by your acquisition cost and check that the money exists: the CAC calculator gives the per-customer number, and the burn rate and runway calculator tells you whether your cash lasts long enough to fund it. A plan whose revenue arrives in month 20 and whose cash ends in month 9 is not a plan.
One more sanity check: divide ending MRR by ending customers. If the resulting ARPA drifts far from today's, your assumptions about new-customer pricing and expansion are quietly repositioning the business up-market or down-market. That may be intended. It should not be accidental.
Where a flat acquisition rate levels off
| Net monthly churn | $1,000 new MRR/mo | $5,000/mo | $10,000/mo | $25,000/mo |
|---|---|---|---|---|
| 0.5% | $200,000 | $1,000,000 | $2,000,000 | $5,000,000 |
| 1.0% | $100,000 | $500,000 | $1,000,000 | $2,500,000 |
| 2.0% | $50,000 | $250,000 | $500,000 | $1,250,000 |
| 3.0% | $33,333 | $166,667 | $333,333 | $833,333 |
| 5.0% | $20,000 | $100,000 | $200,000 | $500,000 |
| 0% or negative | No ceiling | No ceiling | No ceiling | No ceiling |
Halving net churn doubles the ceiling, and so does doubling new MRR — but only one of those two is free to do again next year.
Gross churn, net churn and why the difference matters here
Gross revenue churn is what you lose: cancellations plus downgrades, as a share of opening MRR. Net churn is gross churn minus expansion, and it can be negative. This calculator asks for gross churn and expansion separately, then combines them, because the two are driven by different work — retention and support on one side, upsell and pricing on the other. If you only have a net figure, enter it as churn and leave expansion at zero; the projection will be identical, but you lose the ability to see which lever moves the answer.
Assumptions this model makes, and where they break
- Constant rates. Churn, expansion and ARPA are held flat across the horizon. Real churn is highest in the first few months of a customer's life, so a business adding customers fast has a rising blended churn rate that this model will understate.
- No cohort structure. Every dollar of MRR churns at the same rate regardless of tenure or size. Cohort-based models are materially more accurate over horizons past a year and are worth building once you have two years of data.
- Churn applies before new business. A customer signed in month 5 cannot churn in month 5. Over a long horizon this is a small optimistic bias.
- No seasonality and no contract renewal cycle. If most of your annual contracts renew in one quarter, your churn arrives in lumps rather than smoothly, and a monthly average hides the risk concentration.
- Nothing about price changes or plan migrations. A repricing shows up here only if you fold it into the expansion rate, which mixes two different events.
- MRR is not cash and not recognised revenue. Prepaid annual plans change your cash timing entirely without changing a single number in this forecast.
- Acquisition growth is unconstrained. The model will happily compound new customer adds at 10% a month for three years. Sales capacity, market size and payback economics will not.
Where this model sits among the alternatives
Three forecasting approaches are used in subscription businesses, and they suit different questions.
Simple compounding — MRR × (1 + g)ⁿ — is fine for a one-line sanity check and useless for diagnosis, because it cannot tell you whether growth came from selling more or churning less. The MRR-movement recursion used here is the standard operating model: enough structure to separate the levers, few enough inputs to fill in honestly, and the same four movements your billing system already reports. Cohort models track each month's intake as its own population with its own retention curve, which is the only way to forecast accurately when churn varies strongly with tenure. Move to cohorts when the recursion starts to disagree with your actuals by more than a few percent.
Whichever you use, the forecast is only half the exercise. Convert the ending figure with the ARR calculator for board reporting, feed the growth rate and your margin into the Rule of 40 calculator to check whether the growth is efficient, and reconcile the opening balance with the MRR calculator so the number you are projecting from is the number your billing system agrees with. A forecast built on a mis-normalised starting MRR is wrong from month one, and no amount of modelling sophistication recovers it.
Key terms
- Gross revenue churn
- MRR lost in a month from cancellations and downgrades, as a share of the opening MRR. Always positive.
- Expansion rate
- MRR gained in a month from existing customers paying more, as a share of the opening MRR.
- Retention factor
- 1 − churn + expansion. The multiple your existing book is worth next month before new business. Also called monthly net revenue retention.
- CMGR
- Compound monthly growth rate: the constant monthly rate that reproduces the total growth over n months. (Mₙ/M₀)^(1/n) − 1.
- Steady state
- The MRR level at which monthly churn losses exactly equal monthly additions, so the book stops growing. Equal to new MRR divided by the net churn rate.