Recurring revenue — subscriptions, retainers, maintenance plans, memberships — is the most valuable kind of revenue a small business can build, because next month starts at yesterday's baseline instead of zero. But recurring models run on their own arithmetic, and owners who bring one-time-sale intuition to a recurring business misread everything: growth feels slow when it's compounding, and fine when it's doomed. Four numbers run the whole model.
MRR — monthly recurring revenue. The sum of all active subscriptions' monthly value. A $30/mo product with 200 subscribers: $6,000 MRR. Annual plans count at their monthly rate ($360/yr = $30 MRR). One-time fees don't count — MRR's whole point is measuring the repeating baseline, so keep setup fees and one-offs in a separate bucket or the number stops meaning anything.
Churn — the leak. The percentage of customers (or dollars) lost per month. 200 subscribers, 10 cancel: 5% monthly churn. Its complement is retention (95%). Small-sounding churn numbers are large: 5% monthly compounds to losing roughly half your base over a year (0.95¹² ≈ 0.54). Churn is the number most worth being paranoid about.
New MRR — the faucet. Subscriptions added per month, in dollars.
Net change = new MRR − churned MRR. The only honest growth number. 20 new subscribers while 10 cancel isn't "20 growth"; it's 10.
Here's the non-obvious part. Churn isn't just a drag on growth — it caps the size of the business. As the base grows, churn (a percentage) costs more subscribers each month, while your acquisition (roughly a fixed count per month, set by your marketing) stays flat. The bucket leaks in proportion to how full it is; the faucet doesn't care. Equilibrium arrives when the leak equals the faucet:
Ceiling = new customers per month ÷ monthly churn rate.
Adding 20/month at 5% churn: ceiling = 20 ÷ 0.05 = 400 subscribers ($12,000 MRR at $30). Forever. Not for lack of hustle — that acquisition pace cannot push the base past the point where 5% of it equals 20. The only ways up: more new per month, or less churn. And they are not symmetric: halving churn to 2.5% doubles the ceiling to 800 — and simultaneously doubles average customer lifetime (≈ 1/churn: 20 months → 40), which roughly doubles LTV, which (per the LTV:CAC lesson) justifies spending more to acquire — which raises the faucet too. Retention improvements hit the model three times; acquisition improvements hit it once. This is why mature subscription operators obsess over churn and beginners obsess over new sales.