What a subscription business is actually doing RECUR

Revenue growth is the one figure that cannot tell you whether the business works

Revenue can grow while the product is failing, for as long as new customers arrive faster than old ones leave. That can run for years. The month it stops being true is the month everything stops at once, because the growth was never the business; it was the difference between two much larger numbers moving in opposite directions.

What tells you is the movement underneath, and it is always exactly five things:

start + new + expansion − contraction − churn = end

That identity has to hold to the penny. Where it does not, this reports it rather than smoothing it, because a set of books that does not reconcile is not a set of books, and in practice a month that will not balance means two systems disagree about what a customer is paying, which is worth finding.

The figures that are actually diagnostic

Net revenue retention is what last month's customers are worth this month, ignoring everybody new. Above 100% the business grows with no new customers at all, which is the single most valuable property a subscription business can have and the one an investor will look at first. It is also the number that a headline growth rate can hide completely.

The quick ratio is what you gained for every pound you lost. It answers whether the sales effort is filling a bucket or a sieve, and it moves long before revenue does.

Revenue churn is not customer churn, and the gap between them is the interesting part. Losing a lot of customers and little revenue means the small accounts are leaving, which may be fine. Losing little revenue's worth of customers but a lot of revenue means a large one left, which is not.

Payback is how many months of margin it takes to recover what you spent acquiring a customer. Alongside retention it decides whether growing faster makes you stronger or just emptier.

Getting the inputs honest

Classify the movements consistently, and write the rules down. A customer who downgrades and then leaves in the same month is contraction then churn, not both counted at full value. A customer who left in March and came back in July is new in July, not un-churned in March. Whichever convention you pick, pick one, because the ratios are only comparable month to month if the classification is.

Use one definition of monthly recurring revenue, on one basis, and keep one-off fees out of it. Set-up fees, professional services and usage overages are real money and they are not recurring; counting them flatters expansion and makes retention look like something it is not.

And look at the series rather than the month. One month's retention is noise, especially with few customers; three months of a falling quick ratio is a finding. This works out the movement from the five numbers so you can keep the series honestly rather than recomputing a dashboard's definition each quarter.

Nothing you type is uploaded, which for a revenue series nobody has published yet is the right default. Nearby: CASHFLOW answers the different and more urgent question of when the money is actually in the bank, BREAKEVEN finds the lever worth pulling, LEDGER is the cash book underneath it, and CHART draws the series without uploading it.