Put the last twelve months of cancellations in one sheet
Open a spreadsheet and list every customer who left in the past twelve months. If you have fewer than 20, take the last two years. Use these columns:
- Customer name and the plan or product they had.
- Start date and end date.
- How they found you.
- Size of the account, in revenue or seats.
- The reason they gave, in their own words.
- Whether the cancellation was by choice or came from a failed payment.
Your billing tool, whether Stripe Billing, Chargebee or Recurly, holds the dates and amounts. The reasons live in emails, notes and your head. Write them down now, while you remember them.
Measure churn two ways
Work out logo churn and revenue churn for each month. Logo churn is customers lost divided by customers at the start of the month. If you started with 200 customers and lost 6, that is 3 per cent. Revenue churn is the same sum on revenue.
Do not skip the annual view. A monthly 3 per cent sounds small, but it compounds: after twelve months you have lost about 31 per cent of that starting group. That one calculation usually changes how seriously people take the problem.
Sort every reason into a short list
Group the reasons into about six buckets. These work for most B2B businesses:
- Never a good fit, so they should not have bought.
- Never got started, so onboarding failed.
- Stopped seeing value over time.
- Budget cut or price.
- Moved to a competitor.
- Business closed, merged or changed hands.
Then mark each bucket as preventable or not. Closures and mergers are usually not. The first four usually are. If you find that 40 per cent of cancellations are people who never got started, you know where to work.
Find where the churn clusters
Look for a pattern in three places. First, time: how many months after the start do people leave? A spike in the first 90 days points to onboarding, which is covered in How to design customer onboarding. A spike at the renewal date points to the process in How to design renewal processes.
Second, segment: is one size of company or industry leaving faster? Third, source: do customers from one channel leave sooner? If so, the problem may sit in your sales qualification, not in your product.
Pick one cause and give it an owner
Choose the largest preventable bucket and write one sentence about it: "Customers with fewer than three users at day 30 leave within six months." Give it one owner and one change to try for a quarter. Fixing everything at once means you will not know what worked.
Ask the leaver one question
When someone cancels, ask a single question by email or in a short call: "What was the main reason you decided to stop?" Keep it to fifteen minutes at most and do not argue or offer a discount in that call. You want the truth, not a save.
If a customer is still undecided, treat it as a diagnosis, as described in A save is a fix delivered in person, not a discount thrown in panic.
Catch the problem before the cancellation
Once you know the usual reasons, turn them into warning signs. If never getting started is the main cause, the warning sign is "no login in the first ten days". Build the scoring and triggers using How to monitor customer health and Proactive outreach to at-risk accounts.
Fix the payments that fail by accident
Some churn is not a decision. Cards expire and payments fail. Turn on automatic retries in your billing tool and send two or three clear emails when a payment fails, with a link to update the card. This is often the cheapest churn you can remove.
Win back selectively
Contact former customers 60 to 90 days after they leave, and only if you have fixed the thing they left over. One email, one line on what changed, one link. If nothing changed, do not write.
Common mistakes
- Offering a discount before you know the reason.
- Counting all churn as one number, so failed payments hide among real decisions.
- Taking the reason from the customer's first answer without checking the usage data.
- Running a win-back campaign before the product problem is fixed.
- Reviewing churn once a year, when a monthly check would catch it.
How you know it works
Logo churn and revenue churn fall over two or three quarters, and the preventable buckets shrink while the unavoidable ones stay the same. You can also name the top three reasons customers leave without opening a spreadsheet. If you cannot, run the first step again with this quarter's data.
