MRR & churn
In plain English
MRR means "monthly recurring revenue" — how much money your subscriptions bring in every month, on average.
It's the number that makes subscriptions worth having. It tells you what next month looks like before it starts.
The tricky part is that not everyone pays monthly. Some pay weekly, some quarterly, some yearly. So to get one comparable number, everything is converted to "what this is worth per month":
| They pay | Counts toward monthly MRR as |
|---|---|
| £10 every week | ~£43.33 (there are about 4.33 weeks in a month) |
| £40 every month | £40 |
| £120 every 3 months | £40 |
| £480 once a year | £40 |
All four customers are worth the same per month. That's the whole point of the conversion.
One important detail. MRR here is worked out from what customers actually paid last time — not from your list price. So if someone signed up with a first-order discount, or their plan price changed, the real number is used.
And if a subscriber hasn't been billed even once yet, they count as zero until they are. That may look conservative. It's deliberate: money that hasn't moved isn't revenue.
How to use it
- Open your subscription metrics in the Revenue area of the app.
- Read MRR first — that's your baseline monthly income from subscriptions.
- Check at-risk MRR. That's revenue currently sitting in failed payments.
- Contact those customers — see Subscribers. Most failures are fixable.
- Watch ARPU over time. Rising means each subscriber is worth more; falling with growing subscriber numbers means you're adding cheaper ones.
- Compare MRR month to month. Growing MRR with flat subscriber count means better customers, not just more.
⏱ ~5 min/week · 💳 Pro+ · 🎯 Know next month's income before it starts
Why this matters for your business
Most subscription reporting is quietly wrong, and in a flattering direction.
The common approach is to take your plan price, multiply by subscriber count, and call that MRR. It over-states almost always. Some subscribers are on old pricing. Some came in on a promotional rate. Some signed up yesterday and haven't been charged at all. Some are in dunning and about to stop paying entirely. None of that shows in a price-times-count calculation.
This one starts from the other end: what each subscriber actually paid, last time money moved. That's harder to compute and much harder to argue with. When you plan stock or hiring against it, you're planning against real money.
The at-risk figure is the other half. Failed payments are already tracked in Subscribers; this puts a monthly value on them. "68 failed subscriptions" is a fact. "£2,900 of monthly recurring revenue is currently failing" is a decision — it tells you immediately whether this is worth an afternoon.
Separating those two numbers also fixes a reporting error that catches a lot of teams: counting dunning subscribers inside your MRR. They're still nominally subscribed, so it's tempting. But they aren't paying, and including them means your MRR silently overstates by however much is stuck in dunning.
What this typically unlocks
| What you get | Typical result |
|---|---|
| Accuracy of your recurring-revenue figure | Real, not estimated from list prices |
| Revenue at risk right now | Quantified in monthly terms |
| Ability to plan stock and staffing | Against a number you can trust |
| Spotting quality vs. quantity of growth | ARPU trend makes it obvious |
| Overstated MRR from dunning subscribers | Eliminated — they're counted separately |
What you actually get
The three numbers
| Metric | What it means | What to watch |
|---|---|---|
| MRR | Monthly recurring revenue from active subscribers | The baseline. Should grow steadily |
| At-risk MRR | Monthly value of subscriptions with failed payments | Should be small. If it's climbing, act |
| ARPU | MRR ÷ active subscribers — average revenue per subscriber | Tells you whether growth is quality or quantity |
How each frequency converts
| Their plan | Monthly conversion |
|---|---|
| Every week | × ~4.33 |
| Every 2 weeks | × ~2.17 |
| Every month | × 1 |
| Every 2 months | × 0.5 |
| Every 3 months (quarterly) | × ~0.33 |
| Every year | × ~0.083 |
Worked example. A subscriber paying £15 every 2 weeks:
£15 × (52 ÷ 12 ÷ 2) = £15 × 2.17 = £32.50 MRR
What counts and what doesn't
| Subscriber | Counts toward MRR? |
|---|---|
| Active, has been billed | Yes — at their last actual charge |
| Active, never billed yet | No — counts as zero until money moves |
| Paused | No |
| Failed / in dunning | No — counted in at-risk MRR instead |
| Cancelled | No |
The "never billed yet" rule matters. A subscriber who signed up this morning might cancel before their first charge. Counting them at full value would inflate MRR with revenue that may never arrive. The number stays conservative and correct.
Reading ARPU
| Pattern | What it means |
|---|---|
| MRR ↑, subscribers ↑, ARPU flat | Healthy growth — more of the same |
| MRR ↑, subscribers flat, ARPU ↑ | Better customers, or successful upselling |
| MRR ↑, subscribers ↑↑, ARPU ↓ | Growing on cheaper plans — check margin |
| MRR ↓, subscribers flat, ARPU ↓ | Discounting, downgrades, or plan changes |
Row three is the one to watch. Revenue rising while ARPU falls can mean you're acquiring subscribers who cost the same and are worth less.
Freshness
These figures are calculated periodically rather than recomputed on every page load, so a change may take a few minutes to appear. For anything time-sensitive, the individual subscription states in Subscribers are live.
How it works (without the technical bits)
Real merchant scenarios
Scenario A — Real MRR was 22% below the estimate
Setup. Coffee subscription. The team had been reporting MRR as plan price × subscriber count: 1,180 subscribers × £26 = £30,680.
Real MRR: £23,940.
Where the £6,740 gap came from:
| Cause | Effect |
|---|---|
| 140 subscribers on old £19 pricing | −£980 |
| 96 on a launch promotional rate | −£1,340 |
| 61 signed up but never billed yet | −£1,586 |
| 84 in dunning, not paying | −£2,184 |
| Various part-refunds and adjustments | −£650 |
Why it mattered. They'd committed to a green-coffee purchase based on the inflated figure and had to renegotiate.
Afterwards they planned against real MRR and stopped over-committing.
Scenario B — At-risk MRR justifying an afternoon
Setup. Merchant knew they had "some failed payments" and never prioritised them.
Then they saw the value:
MRR £18,400
At-risk MRR £2,910 (15.8% of MRR)
£2,910 a month — nearly £35,000 a year — sitting in expired cards.
Action. One afternoon: an email to every failing subscriber with an update-payment link.
| Result | |
|---|---|
| Recovered within 2 weeks | £2,140/month |
| Annualised | ~£25,700 |
| Time spent | ~4 hours |
The count alone hadn't motivated anyone. The monthly value did it immediately.
Scenario C — ARPU catching bad growth
Setup. Supplements brand, MRR up 34% over two quarters. Celebrated.
The fuller picture:
| Q1 | Q3 | |
|---|---|---|
| MRR | £22,100 | £29,600 |
| Active subscribers | 610 | 1,140 |
| ARPU | £36.23 | £25.96 |
Subscribers up 87%, MRR up only 34%, ARPU down 28%.
Cause. A heavily-discounted entry plan launched in Q2 had taken over acquisition — and it was also cannibalising the standard plan, with existing customers switching down.
Margin check. The entry plan's contribution margin was 19% against 48% on the standard plan. Gross profit had actually fallen slightly despite MRR growing a third.
Action. Restricted the entry plan to genuinely new customers and raised its price. ARPU recovered to £31.40 over the following quarter with MRR still growing.
On its own, MRR said "great." ARPU said "look closer."
Scenario D — Why zero for un-billed subscribers is right
Setup. Merchant ran a big launch promotion. 340 sign-ups in four days, none yet billed.
They expected MRR to jump immediately. It didn't — those 340 counted as zero.
Their first instinct was that the number was broken.
What happened next. Of the 340, 89 cancelled before their first charge — a normal pattern for promotion-driven sign-ups.
Had MRR counted them at full value, it would have shown a £9,860 jump and then dropped by £2,580 a month later, with no obvious explanation.
Instead MRR rose steadily as real charges landed. The number never told them something that wasn't true.
Best practices
✅ Plan against real MRR, not price × count. Scenario A.
✅ Check at-risk MRR weekly. It converts a vague problem into a decision.
✅ Track ARPU alongside MRR. MRR alone hides quality problems — Scenario C.
✅ Expect MRR to lag a launch. New subscribers count once they've paid.
✅ Keep at-risk MRR under about 5% of MRR. Above that, your payment-recovery process needs attention.
✅ Compare MRR month to month, not week to week. It's a monthly metric.
❌ Don't include dunning subscribers in MRR. They aren't paying. That's what at-risk is for.
❌ Don't celebrate MRR growth without checking ARPU.
❌ Don't estimate MRR from list prices. It over-states, always.
❌ Don't panic if MRR looks lower than you expected. It's probably the first honest number you've had.
Plan tiers
| Capability | Free | Starter | Pro | Agency | Enterprise |
|---|---|---|---|---|---|
| Real MRR from payment history | — | — | ✓ | ✓ | ✓ |
| At-risk (dunning) MRR | — | — | ✓ | ✓ | ✓ |
| ARPU | — | — | ✓ | ✓ | ✓ |
| Cadence normalisation | — | — | ✓ | ✓ | ✓ |
| Billed-in-period revenue | — | — | ✓ | ✓ | ✓ |
| Status mix breakdown | — | — | ✓ | ✓ | ✓ |
| Multi-store MRR roll-up | — | — | — | ✓ | ✓ |
Frequently asked
Why is my MRR lower than price × subscribers? Because it uses what people actually paid. Old pricing, promotional rates, un-billed sign-ups and dunning subscribers all account for the difference — see Scenario A.
Why do new subscribers not show up straight away? They count once they've been charged. Money that hasn't moved isn't revenue — Scenario D.
How is a weekly subscription converted? Multiplied by about 4.33, since there are roughly 4.33 weeks in an average month.
Are paused subscribers in MRR? No. They aren't being billed.
Are dunning subscribers in MRR? No — they're in at-risk MRR, separately, so your headline figure isn't inflated by money that isn't arriving.
What's a healthy at-risk percentage? Under about 5% of MRR. Above that suggests payment recovery needs work.
Why did MRR drop when I didn't lose anyone? Usually subscribers moving into dunning — they leave MRR and appear in at-risk. Check whether the two moved by similar amounts.
How often do these numbers refresh? Periodically rather than on every load. Live per-subscription state is in Subscribers.
See also
- Subscriptions overview — how the module fits together
- Subscribers — acting on the at-risk subscriptions
- Plans — the pricing decisions that drive ARPU
- Revenue Hub — where these metrics appear
- Revenue Leaks — failed payments as recoverable money
- Revenue command center — total revenue context
- Customer intelligence — lifetime value by channel