- Monthly recurring revenue, and the bridge that explains its change.
- Churn, in customers and in revenue.
- Net revenue retention.
- Payback on what a customer costs to win.
- Runway.
This is for subscription businesses. A marketplace or a shop runs on orders and repeat purchases rather than recurring revenue, and its five are different. There are no benchmarks here: a good churn rate depends on your market and your prices, so the first comparison is with your own months.
Let’s take the invented subscription software company from the revenue forecasting page, just after its Series A: €308,000 of monthly recurring revenue from 4,000 customers, €77 each on average, on plans at €29, €99 and €299 a month. It does not exist, and every figure below is made up and computed when this page is built. One month of its numbers runs through all five.
The five, defined
Three of the five come out of the month’s bridge, and they are the forecast’s inputs
1. Monthly recurring revenue, and its bridge
Monthly recurring revenue (MRR) is what every paying customer is contracted to pay per month, summed at the end of the month. An annual plan counts as its price divided by twelve. Discounts come off. One-off fees, services and sales tax stay out, and a customer on a free trial counts nothing. Usage charges that vary from month to month are reported beside it, not inside it.
The usual mistake is to sum invoices instead. A customer who prepays a year then adds the whole year to the month it is invoiced in, as the first check in How to check the numbers in an AI-generated report shows.
The bridge explains the month’s change in four lines. MRR at the start, plus new customers, plus expansion (existing customers paying more), minus contraction (paying less), minus churned (customers who left), equals MRR at the end. If it does not add up, a line is missing: often customers who came back, or a price change applied to everyone. The invented company has customers who come back, so its bridge has that fifth line.
New and returning customers carried the month: without them, recurring revenue would have fallen €4,124
The same data as a table
| Line | Customers | Monthly recurring revenue |
|---|---|---|
| Start of the month | 4,000 | €308,000 |
| New | +150 | +€4,350 |
| Came back | +34 | +€986 |
| Expansion | +€5,760 | |
| Contraction | −€1,640 | |
| Churned | −156 | −€8,244 |
| End of the month | 4,028 | €309,212 |
MRR grew €1,212, or 0.4%. All of that growth, and more, came from customers who started paying in the month, €4,350 from new ones and €986 from those who came back: without them, MRR would have fallen €4,124.
2. Churn, in customers and in revenue
Customer churn is the customers who stopped paying during the month, divided by the customers paying at its start. Revenue churn is the MRR lost to cancellations and downgrades, divided by MRR at the start. Both take the start of the month as the base, so a customer who joins and leaves within the month is in neither; count those separately.
In the invented month, 156 of 4,000 customers left: 3.9%. They took €8,244, and downgrades took another €1,640, so revenue churn was €9,884 on €308,000, or 3.2%. The leavers paid €53 a month on average, against €77 across the base: the company is losing smaller-than-average customers. Revenue churn below customer churn is the better of the two patterns. The reverse means you are losing a larger share of revenue than of customers: the ones leaving paid more than average, or the ones staying cut back.
The usual mistake is a base that changes from month to month, such as every customer active at any point in the month instead of those paying at its start. A promotion that brings in a wave of short-lived customers then lowers churn on paper, as the denominator check shows.
3. Net revenue retention
Net revenue retention (NRR) compares what a fixed group of customers pays now with what the same customers paid at the start of the period. Expansion counts, contraction and churn come off, and every customer won since is left out. Measured over twelve months, it shows what a year does to the revenue of the customers you have; month on month, it is what the example monthly report prints. Gross revenue retention is the same without expansion, so it can never pass 100%.
In the invented month, existing customers went from €308,000 to €303,876: a monthly NRR of 98.66%. One month like that looks harmless. Twelve of them compound to 0.986612 = 85.1%, so a year from now these customers would pay about 15% less, and gross revenue retention would be 67.6%.
Twelve months at 98.66% compound to 85.1%, and to 67.6% without expansion
The same data as a table
| Month | Net revenue retention | Gross revenue retention |
|---|---|---|
| 0 | 100.0% | 100.0% |
| 1 | 98.7% | 96.8% |
| 2 | 97.3% | 93.7% |
| 3 | 96.0% | 90.7% |
| 4 | 94.8% | 87.8% |
| 5 | 93.5% | 85.0% |
| 6 | 92.2% | 82.2% |
| 7 | 91.0% | 79.6% |
| 8 | 89.8% | 77.0% |
| 9 | 88.6% | 74.6% |
| 10 | 87.4% | 72.2% |
| 11 | 86.2% | 69.9% |
| 12 | 85.1% | 67.6% |
The usual mistake is to read NRR above 100% as “nobody leaves”. Expansion from the customers who stay can cover heavy churn from the ones who go, which is why I would always report gross retention beside it.
4. Payback on what a customer costs to win
Customer acquisition cost (CAC) is sales and marketing spend in a period divided by the paying customers won in it. Payback is CAC divided by the margin a new customer brings each month: their MRR times the gross margin. Gross margin is revenue less the cost of serving customers (hosting, support, payment fees, the software inside the product), as a share of revenue.
The invented company spent €47,400 on sales and marketing in the month and won 150 customers: €316 each. At €29 a month on its Starter plan and a gross margin of 80%, a new customer brings €23.20 a month, so payback takes about 14 months. That assumes the customer stays for those months. If new customers leave at Starter’s rate of 5.0% a month, about half of them are gone before the 14 months are up.
By the month a new customer pays back, about half of them have left
The same data as a table
| Month | Still paying | Margin earned by one who stays |
|---|---|---|
| 0 | 100.0% | €0.00 |
| 3 | 85.7% | €69.60 |
| 6 | 73.5% | €139.20 |
| 9 | 63.0% | €208.80 |
| 12 | 54.0% | €278.40 |
| 14 | 48.8% | €324.80 |
| 18 | 39.7% | €417.60 |
| 24 | 29.2% | €556.80 |
The definition needs three choices written down: which costs count (sales salaries, advertising, tools, the founders’ own selling time), whether spend is matched to customers won in the same month or a later one, and whether a customer means a paying one or a trial.
5. Runway
Runway is cash in the bank divided by monthly net burn, which is cash out less cash in. Take the burn as an average of the last three months, so that one large bill or one early payment does not move it. The invented company holds €6.3 million and burned €165,000, €175,000 and €185,000 in its last three months: €175,000 on average, 36 months of runway. If the burn keeps rising €10,000 a month, as it did over those three, the cash lasts about 22 months, not 36.
If the burn keeps rising €10,000 a month, the cash lasts about 22 months, not 36
The same data as a table
| Month | Burn flat | Burn rising |
|---|---|---|
| 0 | €6,300,000 | €6,300,000 |
| 6 | €5,250,000 | €4,980,000 |
| 12 | €4,200,000 | €3,300,000 |
| 18 | €3,150,000 | €1,260,000 |
| 21 | €2,625,000 | €105,000 |
| 22 | €2,450,000 | Run out |
| 24 | €2,100,000 | Run out |
| 30 | €1,050,000 | Run out |
| 36 | €0 | Run out |
Write each definition down
Each name above has more than one reasonable calculation. For listed companies, the US Securities and Exchange Commission’s 2020 guidance generally expects, depending on the facts and circumstances, a metric reported in the management discussion of their filings to come with a clear definition and how it is calculated. When the calculation changes, it asks them to consider disclosing, where material, the difference, the reason and the effect on the figures reported for the current period and earlier ones, and recasting the earlier figures where needed.
I would hold a Series A board report to the same standard. It costs one page of definitions, changed only with a note saying what changed, and it stops a churn rate from improving because someone moved its denominator.
What the five look like each month
Tracked monthly, the five belong in one report, with the bridge as a chart. The example monthly revenue report shows the first three for the same invented company a year on, the bridge as a chart, with the checks that ran on it. The same parts drive the forecast: next month’s MRR is this month’s, less what leaves, plus what existing customers add, plus new customers, which is how revenue forecasting builds it.
How FinDaS does it
FinDaS builds the monthly report and the forecast behind it as a done-for-you service. AI agents write the queries and the checks, automated checks and a review round check every report, and each report carries a short note of which checks ran. Before this, I built reporting functions from scratch, and worked on budgeting and forecasting, during eight years in retail-banking and consumer-finance CRM analytics. What you act on, such as a model going live or a decision memo, goes to a panel of reviewer agents, and then I sign it; the monthly refresh of a signed-off report runs on the agents’ checks. The steps are on how it works.
Sources
- US Securities and Exchange Commission, Commission Guidance on Management’s Discussion and Analysis of Financial Condition and Results of Operations, Release Nos. 33-10751, 34-88094 and FR-87, effective 25 February 2020.