Two kinds of churn

Churn is the rate at which you lose revenue or customers. The important split is why a customer left, because each cause needs a different fix.

• Involuntary churn. The customer wanted to stay but a payment failed. Expired cards, insufficient funds, and bank declines all land here. You fix it with recovery emails and retries.
• Voluntary churn. The customer decided to leave. You fix it with a better cancel experience, a well-timed offer, and learning the reasons behind it.

Adding both together can hide the real problem, so churnguardian always splits them apart.

Dunning

Dunning is the process of recovering a payment after it fails, or the process of managing quiet hours and reminders. churnguardian runs a short sequence of emails on a schedule:

• Day 0. A friendly heads-up right after the charge fails, with a one-click card update link.
• Day 3. A reminder that keeps access top of mind.
• Day 7. A final notice before the account is paused.

The moment the payment recovers, the sequence stops, so no one gets an email they do not need.

Smart retries

Many failures are temporary. A card can decline because of insufficient funds one day and clear a few days later. churnguardian reads the decline code and retries on a schedule that fits it:

Cancel flows

A cancel flow is a hosted page a customer sees when they try to cancel. It does three things:

• Asks why they are leaving, so you capture the reason.
• Shows an offer, such as a discount or a pause, chosen for them.
• Either applies the offer and keeps the plan active, or lets them confirm the cancellation.

Because the flow lives on its own page, it needs no code on your cancel button. You just redirect the customer there.

Adaptive offers

A static offer gives everyone the same discount. An adaptive offer learns. For every plan and billing interval, churnguardian tests combinations of discount and duration, then serves the ones that save the most customers. This is a form of multi-armed bandit, a small amount of learning that favors what works while still exploring.

To keep the learning honest, a small slice of customers never sees it:

• The test arm. About 85 percent of customers see the offer chosen by the bandit.
• The control arm. The other 15 percent always see your fixed offer. Their results are the baseline every adaptive number is measured against.

Measuring recovery

These are the numbers you will see most often: