What is churn for an ISP, and how do you measure it?
Churn is the share of subscribers who stop paying you over a period. For an internet provider the standard measure is monthly churn rate: subscribers lost during the month divided by subscribers at the start of the month. Count disconnects, not new sales net of losses, or growth will hide the problem.
| Metric | How to calculate it | Why it matters |
|---|---|---|
| Monthly churn rate | Subscribers lost in the month ÷ subscribers at the start of the month | The headline number to track every month |
| Voluntary churn | Subscribers who chose to cancel or port out | Driven by service, price and competition |
| Involuntary churn | Subscribers disconnected for non-payment and never recovered | Often the easiest churn to win back |
| Revenue churn | Monthly recurring revenue lost, including downgrades | Captures value lost, not just headcount |
Split churn into voluntary and involuntary from the start. They have different causes and different fixes. Track revenue churn alongside it too: a subscriber who downgrades still lowers ARPU, and downgrades are often the step before a cancellation.
Why do subscribers leave?
The reasons are rarely a mystery. What makes them hard to act on is that they are spread across different systems and different teams.
- Service quality
- Outages, slow evening speeds and repeat faults in the same service area. The subscriber experiences it every day.
- Billing surprises
- An unexpected charge, a promotional price rolling off, or an invoice line nobody can explain.
- Price
- A competitor offer that looks better on paper, especially at contract end.
- Competitor entry
- A new fiber build or fixed wireless network arriving in your footprint.
- Support experience
- Long holds, being asked to repeat themselves, or a problem that took several calls to fix.
- Moves
- Relocation outside your coverage. Mostly unavoidable, but worth separating in your reporting.
Most of these leave a trace well before the cancellation. The subscriber called about speeds, paid late twice, or sat in a sector with repeated faults. The question is whether anyone connects those traces in time.
Which early-warning signals are already in your data?
You don’t need new data to predict churn. You need to read the data you already have in one place, per subscriber, every day.
| Signal | Where it lives | What it tends to predict |
|---|---|---|
| Repeat tickets for the same issue | Help desk | Frustration with an unresolved problem |
| Outage exposure by sector | Network monitoring, OLT or tower data | Service-quality churn in a whole area |
| Slow-speed complaints at peak hours | Calls and tickets | Capacity problems the subscriber feels daily |
| Late or failed payments | Billing | Involuntary churn, and sometimes intent to leave |
| Plan downgrades | Billing | Price sensitivity; often the step before cancelling |
| Negative sentiment on calls | Call recordings and transcripts | Intent that never reaches a ticket |
Signals are strongest in combination. One slow-speed complaint means little; a complaint, a sector fault and a late payment in the same month is a subscriber worth calling. Clustering issues by theme and service location shows when a problem is systemic rather than individual:
Zone 07 flagged. Ninety-four contacts from one service area in nine days, clustered before a single one was filed as an outage.
How do you fix involuntary churn?
Involuntary churn is subscribers you lose to non-payment, and a large part of it is recoverable. Many of these people want to keep the service; the payment simply failed, or arrived at a bad time of the month. The fix is a dunning process that makes paying easier than leaving.
- 1.Remind before the due date, not only after it, in the channel the subscriber actually reads.
- 2.Make payment one step: a pay-by-text link or a saved payment method, not a portal login.
- 3.Offer a payment plan before suspension, split into amounts the subscriber can manage.
- 4.Suspend on clear, published rules, and reconnect the moment payment is confirmed, at any hour.
- 5.Follow up with subscribers suspended for more than a few days before they become a disconnect.
- DueSep 01
- ReminderSep 08
- SuspendedSep 15
- Paid via linkSep 18 14:02
- ReconnectedSep 18 14:04
Automatic reconnection matters more than it looks. A subscriber who pays at 9 PM and waits until the next business day for service is a subscriber who starts comparing providers. See how the billing and collections agent handles reminders, payment links and reconnection.
Which retention offers work, and when should you make them?
Timing matters more than the size of the offer. An offer made weeks before a subscriber decides to leave feels like service; the same offer made during the cancellation call feels like a bargain they had to fight for, and it costs more.
- 1.Fix the cause first. If the subscriber is at risk because of repeated slow speeds, a discount without a fix only delays the cancellation.
- 2.Match the offer to the signal. A price lock suits price sensitivity; a speed upgrade suits a capacity complaint; a credit suits an outage.
- 3.Set limits in advance. Decide which offers each risk level can receive, so offers are consistent and margins are protected.
- 4.Reach out in the subscriber’s channel. Some answer a call, others only answer a text.
- 5.Confirm in the same conversation. Every extra step between yes and applied loses saves.
Don’t give every at-risk subscriber the same discount. Blanket offers train your base to threaten cancellation, and they spend budget on subscribers who were never going to leave.
How do you measure whether retention is working?
Measure retention work against what would have happened without it, not against last year.
| What to track | Why |
|---|---|
Churn rate, voluntary and involuntary, by month The outcome itself, split by cause. | The outcome itself, split by cause. |
Save rate among contacted at-risk subscribers Shows whether outreach changes decisions. | Shows whether outreach changes decisions. |
Churn among at-risk subscribers you did not contact The baseline your saves are measured against. | The baseline your saves are measured against. |
Cost of offers per retained subscriber Keeps retention from quietly eroding ARPU. | Keeps retention from quietly eroding ARPU. |
Retained subscribers still active a few months later Separates real saves from delayed cancellations. | Separates real saves from delayed cancellations. |
Recovery rate of suspended accounts The clearest measure of involuntary churn work. | The clearest measure of involuntary churn work. |
Where does Inerxia fit?
Inerxia’s retention agent scores every subscriber daily on usage, billing history, network quality and sentiment, then intervenes weeks before the cancellation with the offer and channel you configure, over outbound voice, SMS and email, with a handoff to your retention desk.
Subscriber Intelligence reads every call and ticket and clusters issues by theme and service location, so a sector fault shows up before it turns into cancellations. The billing and collections agent works involuntary churn: reminders, payment plans and your own payment links inside the conversation, with service restored automatically once your processor confirms the payment.