A subscription business is a bucket with a tap and a leak. The metrics describe the tap, the leak, and the value of what is in the bucket.
Revenue: monthly recurring revenue, split into new, expansion, contraction, and churned MRR. The split is the point; a flat MRR built from strong new sales and heavy churn is a sick business that looks healthy.
The leak: logo churn (customers lost) and revenue churn (money lost), monthly, by cohort. Net revenue retention above 100% means the existing base grows on its own; that is the single healthiest sign.
The tap: new customers, CAC by channel, and payback period (months of gross margin to recover CAC). LTV to CAC around 3 or better, and payback under 12 months for a small business, are the usual bars.
Leading indicators: activation rate (reached the first value moment within the first week), engagement (weekly active share of subscribers), and support load. Churn is a lagging metric; activation and engagement predict it.
Then the cohort view: retention curves by signup month. A curve that flattens means a core of customers who stay; a curve that keeps falling means the product has not found its keepers, and no amount of acquisition fixes that.
Worked line: "MRR grew 8% but net revenue retention is 85% and the payback period stretched from 9 to 15 months. Growth is being bought. The engagement metric that leads churn dropped two months ago; that is where to look."
What they are checking: the MRR decomposition and the difference between lagging and leading metrics.
Common mistake: reporting total MRR and total subscribers with no churn split, which hides exactly the thing the question is about.