Skip to main content
GBrain
Menu
Log in Get Started

Churn Optimization

Seven skills that ask every customer who leaves what went wrong, learn the warning signs from what they say and do, calculate every Monday who may leave next, and check in with them before they decide.

What comes in, what it makes, where it lands

Tap a skill to see what it reads and makes.

  1. When a customer leaves, ask them why while they still read your email.

  2. Every week, turn what they said and did into warning signs you can count.

  3. Before the next one leaves, act on it: with the customer, and in the product.

    • You build for what makes customers stay

What it found on GBrain's own customers

  • 55% vs 18%

    Customers who used GBrain on two days or fewer in their first two weeks cancelled 55% of the time. Everyone else, 18%.

    So we build for a second day of use.

  • 21 of 55

    customers who left got stuck setting it up. Most of them picked "too expensive" on the cancel form.

    So the fix was setup, not a discount.

  • Every one

    of the customers who asked what it costs stayed.

    So a pricing question is a good sign, and gets a fast answer.

The only emails are the short notes to customers leaving and likely to leave, and none goes out until you turn on auto-send. Each Monday's list is checked later against who really left.

Know who is about to cancel while there is still time to help, and why the last ones did.

Seven skills, one churn file

Stripe tells you who left. It doesn’t tell you why, or who is next. The why is spread across your calls, your email, the cancel form and what the customer actually did in your product. Churn is seven skills that ask, read all of it, keep what they read, and turn it into one list every Monday: the customers most likely to leave, with the reasons in plain words.

It starts by asking. Cancel Emails writes a short, honest note to each customer who sets their plan to cancel: sorry, thank you, would you tell us what went wrong. Failed Payment Emails writes the same note when a payment fails, because a cancelled card is often a customer leaving without saying so. Most people answer.

Then it reads. Why They Leave writes a full story for each customer who stopped paying: what they wanted, what they told you (their answer to the note included), what they did week by week, and the one main reason they left. Why They Stay writes the same story for customers who have paid for months, so there is something to compare against.

Then it learns. Warning Signs finds what the customers who left did differently in their first two weeks, tests each pattern on every past customer, keeps the ones that really come before a cancel, and drops the ones that don’t. Churn Risk calculates who may leave next: every Monday, each paying customer is likely to leave, worth watching, or fine, with the reasons and what they pay.

Then it checks in. Check-in Emails writes one short personal note to each customer likely to leave, every Monday, matched to why they were flagged: an offer to set it up together, “anything get in the way?”, or one thing to try. It holds one in four back, so eight weeks later you can see whether the notes helped.

These notes are the only emails. None goes out until the sender turns on auto-send, and no customer gets a check-in once they have set their plan to cancel.

Each skill is a page of plain instructions in your workspace. Read it, change it in chat (“count a customer as old enough after 30 days, not 45”, “leave out our own team’s accounts”), and it says what it changed.

Turn one on, then the rest

Each one sets itself up in a chat: it asks where your customers pay, write to you and use your product, sends you to connect those accounts, and runs. Turn on a second one and it finds the churn file the first one made.

Start with Cancel Emails