पाठशाला Pathshala · ग्राहक Grāhak, The customer · Lesson 14 · Build
The churn interview: learning from the ones who left
The cancel form says “too expensive” and is usually wrong. A twenty-minute conversation within a fortnight of leaving finds the real reason, and a monthly review puts it on the roadmap.
Pathshala, The Founder Library · 11 October 2026 · 7 min read
A customer who leaves has just made the decision every other customer is quietly weighing. Most companies learn nothing from it, because the only record is a dropdown on a cancel form that says the product was too expensive.
This lesson sets out how to run the exit conversation: whom to call and when, how to separate payment failures from real departures, the two stories to ask for, a tree that classifies the real reason, and the monthly review that turns the reasons into work within thirty days.
What the cancel form hides
Exit surveys fail for the same reason most surveys do: they ask the customer to summarise a decision in a word, and the word they choose is the one least likely to start an argument. “Price” is polite. It does not accuse the product, it does not admit they never set it up, and it ends the conversation. Sometimes it is true. Often the real story is that nobody in their office ever learned to use the thing, or that a competitor solved one specific problem on one specific day.
Bob Moesta, who with Chris Spiek developed the switch interview, describes the leaving customer differently. In a 2024 talk at Business of Software he says that a customer leaving you is progress for them, and that an interview with someone who left gives two stories: why they came, and why they left. Both are worth having. The first tells you what you were hired for. The second tells you what failed to happen.
Separate the payment failures first
Before calling anyone, take every cancellation in the month and check the billing record. Some customers did not leave; their payment did. This involuntary churn looks identical on a dashboard and needs a completely different fix.
In India it is common because recurring payments carry more steps. Under the Reserve Bank’s e-mandate rules, as Stripe’s documentation summarises them, a recurring card or UPI debit needs a mandate registered with additional authentication, a notice to the customer at least 24 hours before each debit with the option to opt out, and fresh authentication for any recurring transaction above ₹15,000. The Reserve Bank raised that ceiling to ₹1 lakh in its circular of 12 December 2023, but only for mutual fund subscriptions, insurance premiums and credit card bills; software subscriptions stay at ₹15,000. Checked October 2026.
Each step is a place for a customer to drop out without meaning to: a card replaced, a pre-debit notice ignored, an authentication link that expired. Automatic retries do not rescue many of them. Stripe’s Smart Retries, whose recommended default elsewhere is eight attempts within two weeks, does not retry payments on India-issued cards at all. The fix is human: a call or WhatsApp message within 48 hours, a payment link to re-register the mandate, and a separate line for involuntary churn in every report so it does not hide inside the real thing.
Whom to call, and how soon
Call every customer who chose to leave, within fourteen days of the cancellation. After a month the story has been retold so often it has become the polite version. If more than twenty leave in a month, call a random twenty, not the twenty most likely to pick up; the friendliest ex-customers are the least representative.
The call should come from someone senior who had no part in the account, ideally a founder. Make the ask honest and short: you are not trying to win them back, you want to understand what happened, and twenty minutes would help. Say no save offer will follow, and keep the promise; a churn interview that ends in a discount teaches every future leaver to exaggerate. Offer a small thank-you, such as a gift voucher, after the call rather than before it.
Some will not take the call. Send those a WhatsApp message with three questions they can answer in a voice note: when they first thought of leaving, what they use now and what would have kept them. A voice note in the customer’s own language carries more of the story than a typed form, and owners of small businesses who will not book twenty minutes will often talk for three while driving. Log the answers on the same ledger, marked as partial, and never count a partial answer as the reason for a whole segment. If fewer than half of a month’s leavers give any story at all, the problem is the ask, not the customers: change who calls, and how soon.
The interview: two stories on a timeline
Ask for the arrival story first; it is easier and it warms the conversation. When did they first think they needed something like this? What were they using before? What happened on the day they signed up? Then the departure, walked as a timeline. The switch interview, as Moesta and Spiek’s site describes it, is a forensic, backward walk through a real decision: a first thought, a period of passive looking, an event, active looking, another event, the decision.
Useful questions follow that order. When did you first think this might not be working? What was happening that week? What did you try before deciding to leave? Did anyone in your office still use it at the end? What are you using now, and how did you find it? What happened on the day you cancelled, and who made the call? Then ask the question the cancel form asked, and compare the answer with the story you just heard.
Listen for the four forces the same site names: the push that started the change, the pull of the new option, the anxiety that made them hesitate and the habit they had to break. A customer who left for a competitor felt a strong pull. A customer who went back to Excel felt a weak push and a strong habit. They need different responses, and only the story tells them apart.
Classify the real reason
After each call, walk the story through the tree below before writing anything else. The order is deliberate: payment, then activation, then the job, then what replaced you, then what tipped it. Teams that start with “what does the competitor have that we do not” misclassify every customer who never got set up.
Record the verdict on a churn ledger, one row per departure: date, segment, months as a customer, monthly revenue, verdict, the tipping event in one sentence and one verbatim quote. The revenue column is what makes the ledger argue. Ten customers lost to onboarding at ₹1,500 a month and two lost to a missing integration at ₹25,000 a month are different priorities, and only the rupees show which.
The cancel form records the reason that ends the conversation. The interview records the reason that ended the relationship.
A worked example: a clinic software company in Pune
A company sells appointment and billing software to small clinics at ₹2,500 a month. In September eighteen of four hundred customers cancel, ₹45,000 of monthly revenue, and the cancel form says “price” for eleven of them. The founder checks billing first: seven are payment failures, mostly mandates that lapsed when a doctor replaced a card. Calls and payment links recover five within a week.
She interviews the remaining eleven. Five never sent a single patient reminder; the receptionist was never trained and the doctor gave up. Two clinics closed or merged. Three moved to a competitor, and all three tell the same tipping story: a patient complained about a missed appointment, and the competitor sends reminders on WhatsApp. One left over a price rise. The ledger reads: onboarding ₹12,500 a month lost, WhatsApp reminders ₹7,500, job ended ₹5,000, price ₹2,500. “Price” was the stated reason for eleven and the real reason for one.
Within thirty days the company assigns onboarding an owner and adds a training call with the receptionist in week one, and puts WhatsApp reminders on the quarter’s roadmap. The October ledger is the test of both.
The monthly churn review
Forty-five minutes in the first week of every month. Read the ledger: involuntary first, recovered and lost, then each verdict with its revenue. Read five quotes aloud. Pick the top reason by revenue lost and give it an owner, one change and a date within thirty days. Check last month’s change: did the reason it targeted shrink in this month’s ledger? Track churn by reason as a line, not churn as one number; the [retention lesson](/library/retention-curves-and-flattening-test) shows how to read it by cohort.
Keep the calls going when churn falls. A month with four departures is a month with four stories nobody else in the market is telling you.
Payment rules quoted here were checked in October 2026 and change; confirm the current limits with your payment provider before relying on them.
Sources
- Bob Moesta, Live Jobs to be Done Case Studies and Problem Shooting, Business of Software Europe, 2024 (talk and transcript) — Leaving is progress for the customer; a churn interview yields the story of why they came and why they left.
- Jobs to be Done (Bob Moesta and Chris Spiek), The switch interview — Timeline from first thought to purchase; the four forces of push, pull, anxiety and habit.
- Stripe Docs, India recurring payments (RBI e-mandate requirements) — Mandate with AFA, pre-debit notice at least 24 hours ahead, AFA each time above ₹15,000. Checked October 2026.
- Reserve Bank of India, Processing of e-mandates for recurring transactions, circular of 12 December 2023 — Limit raised to ₹1 lakh only for mutual fund subscriptions, insurance premiums and credit card bills.
- Stripe Docs, Automate payment retries (Smart Retries) — Recommended default of 8 tries within 2 weeks; no automatic retries for India-issued cards. Checked October 2026.