How I Identify Customer Risks Before They Become Problems

To identify customer risks early, watch how three things change over time: product usage, support tickets and how fast the customer replies to you. When two of them move the wrong way at the same time, call the customer. One warning sign is a reason to look closer. A combination is a reason to act.

What are the early signs a customer is at risk?

Customers rarely leave overnight. In the months before a cancellation there are usually small changes. These are the signals I check, with the starting thresholds I use. They are my own rules of thumb, not industry benchmarks, so adjust them to your product.

Signal	What to look for	My starting threshold
Usage	Logins or key actions falling	Down 30% or more vs the previous 4 weeks
Support tickets	Sudden rise for that account	Double the usual monthly count
Reply speed	Contact answers slower than before	Same-day replies turning into a week of silence
Main contact	Your champion changes role or leaves	Any change, always
Meetings	Check-ins cancelled or skipped	Two in a row
Payment	Invoices paid late	First late payment after a clean history

The change matters more than the absolute number. An account that always logs in twice a week is fine. An account that went from ten logins to two is not.

Why is one signal not enough?

Single signals create false alarms. Compare three accounts:

• Account A: usage down 10%. Probably nothing.
• Account B: usage down 40% and tickets doubled. Worth a look this week.
• Account C: usage down 40%, tickets doubled, and no reply for two weeks. Call now.

Context counts too. A usage drop in August may simply be the summer holiday in France, and a drop during Ramadan may reflect shorter working hours with partners in the Gulf. Compare with the same period last year before you raise the alarm.

The partner I nearly lost

The first partner I nearly lost showed only one clear sign. A hotel group in North Africa that always replied the same day went quiet for two weeks before renewal. Usage looked normal, so nobody worried. When I finally called, my contact had left, and the new manager had never heard of us. We kept the account, but only because the call happened before the renewal date. Since then, a silent champion counts as a risk on its own.

How do I turn risk signals into one score?

Once you know which signals matter, weight them into a single health score so you can sort accounts in seconds. The health score calculator lets you test weights on your own accounts, and the customer health score model shows a full setup. Check the score against reality: look at last year's churned customers and see whether it would have flagged them.

What do I say when I spot a risk?

Keep the first message short and specific. "We noticed your team's usage changed over the last few weeks. Is something getting in the way?" works better than a general check-in. The dashboard tells you something changed. Only the conversation tells you why.

Try this this week

1. List your ten largest accounts.
2. For each, note the usage trend, the ticket trend and the date of their last reply.
3. Mark any account where two of the three are moving the wrong way.
4. Call those accounts within two weeks, starting with the nearest renewal.
5. Write down what you learned on each call, so next month's thresholds get better.

To put a number on what those accounts would cost if they left, run your figures through the churn rate calculator. The previous edition on turning complaints into improvements covers the ticket side in more detail.