Atria • Free operator guide Your subscription app shows you snapshots: active subscribers today, MRR today, churn this month. What it rarely shows is the movement underneath, and that movement is where the money goes missing.
This guide covers the seven numbers worth tracking, the formula behind each one, the benchmarks to compare against, seven fixes that actually move the numbers, a ten-minute weekly routine, and a prompt you can paste into an AI tool to build your dashboard from a single export.
A single blended cancel rate mixes together three problems that have nothing in common, and each one needs a different fix.
Cancels after order 2 are an ad problem. The customer never got what the ad promised them, or had no idea when to expect results. This is the most common leak and the most fixable one, and we come back to it at the end of this guide.
Cancels after order 5 are a product problem. They did get the result, and then it stopped feeling worth the money every month. The fix lives in your term structure, flexibility and product range.
A declined card isn’t a cancel at all. Most stores count it as one anyway and never try to get the money back. The fix is payment retries and a one-click card update.
A healthy-looking average can hide a cliff. A 7% monthly churn rate can be a steep drop before the second order sitting on top of a small, steady leak from loyal subscribers. One number, two problems, opposite fixes.
MRR × 12 is not next year’s revenue. Ecommerce subscriptions churn far faster than software. At 7% monthly churn, the next twelve months collect roughly MRR × 7.7, about a third less than the MRR $\times 12$ headline. Plan inventory or a fundraise on the headline and you will come up short.
A monthly rate hides the compounding. Start with 1,000 new subscribers. Lose 35% before order 2 and you have 650. Lose another quarter of those before order 3 and you are down to about 488. Less than half the group is left after three shipments, and you paid to acquire all 1,000. The answer to all three is the same: track churn by order number, separate failed payments from real cancels, and forecast with your actual churn rate instead of multiplying by twelve.
Active subscribers. Live contracts, excluding cancelled and expired. Decide once whether paused subscriptions count as active, then never change the rule.
Gross MRR. Every active contract converted to a monthly value, with 8-week, quarterly and annual plans all normalized, then added up. This is your headline number.
Live MRR. Gross MRR minus the value of paused and skipped subscriptions. This is what you will actually bill.
Customer churn. Cancels this month divided by subscribers at the start of the month. It tells you how many people leave.
Revenue churn. MRR lost this month divided by MRR at the start of the month. It tells you how much money leaves, and it is the one to watch more closely.
Cohort retention. The share of each signup month still active at order 2, 3, 6 and 12. Follow one signup month across to see where that group falls away. Compare different months at the same order to see whether a change you made actually worked.
Projected revenue. Orders scheduled for the next 30, 60 and 90 days, multiplied by your real collection rate. This is the cash you can plan around.
Five supporting numbers are worth keeping alongside them: the MRR waterfall (new, expansion, contraction and churned MRR, so growth and leakage show up separately), voluntary versus involuntary churn, cancel-flow save rate, payment recovery rate, and churn by order number.