Every sales team celebrates a signed contract. Very few go back six months later to look at that same customer asking to cancel. And that's where an expensive problem hides: if your prospecting keeps targeting the same type of company that leaves quickly, your SDRs are pouring water into a leaky bucket. The CAC has been paid, onboarding has eaten up hours of your CS team's time, and the revenue disappears before it ever covers the acquisition cost.
Churn is the most honest data you have about your ICP. A customer who canceled has already answered, with real money, the question no market research answers well: "Did this company actually need your product?" This article shows how to turn that answer into practical criteria for who not to prospect.
Why ICPs are usually built from the wrong side
Most ICPs start by looking at your best customers: industry, company size, buyer title. That makes sense, but it's only half the job. An inclusion-only ICP produces lists that are far too broad, and SDRs end up reaching out to accounts that look like great customers on paper but share invisible traits with the ones who churned.
Two 200-person companies in the same industry can end up in completely different places. One has a structured sales process and a manager who will actually use the tool; the other bought on impulse because of a VP who left three months later. A firmographic filter can't tell them apart. Churn analysis can.
How to run a churn analysis
You don't need a data team. A spreadsheet with the customers who canceled over the past 12 to 24 months, compared against those who renewed, gets you most of the way there. For each account, capture:
- Time to cancellation: customers who leave within six months usually point to a sales mistake, not a product problem.
- Stated reason vs. real reason: "budget" often means "nobody ever used it." Talk to CS, don't just pull from the CRM.
- Lead source: outbound, inbound, referral, event. Check whether any channel is overrepresented in churn.
- Who bought and who used it: the decision-maker's title, the main user's title, and whether that person was still at the company.
- Account context at the time of purchase: maturity stage, team size in that function, tools already in place, financial situation.
- Discount given: aggressive discounts to close a deal often correlate with early exits.
Then look for patterns that show up frequently among churned accounts and rarely among retained ones. With small datasets, don't expect statistical significance; look for recurring signals and validate them with the people who managed those accounts.
The most common patterns
Every business has its own, but a few come up again and again in B2B tech companies:
Too small to feel the pain
The problem your product solves only becomes painful past a certain volume: transactions, sales reps, patients, deliveries. Below that threshold, customers buy on aspiration and cancel once they realize a spreadsheet would have done the job.
Buyer with no operational owner
When the person who signs isn't the person who'll run it, and there's no clear owner for implementation, the risk of abandonment rises sharply. Treat this as a qualification criterion, not just an account exclusion.
Not mature enough
An advanced analytics tool sold to a company without organized data turns into a consulting project nobody signed up for. The customer doesn't fail for lack of interest, but for lack of foundation.
Wrong timing
Companies going through restructuring, recently acquired, or migrating ERPs rarely have the bandwidth to adopt a new tool. These triggers can be tracked before you ever reach out.
Pro tip: turn each churn pattern into a clear-cut question your SDR can answer before or during the first conversation. For example: "How many active sales reps do you have today?" or "Who would own the implementation?" If the answer falls into the risk zone, the meeting doesn't get booked as qualified, no matter how interested the prospect seems.
From diagnosis to prospect list
Finding the patterns is pointless if they never reach your SDRs' day-to-day work. Three changes make the difference:
- Explicit exclusion criteria in your ICP document. A "do not prospect" section with size ranges, industries, technologies, and situations. Written down, not implied.
- Filters applied when building lists. Anything you can filter with public data (headcount, open roles, tech stack, restructuring news) comes off the list before the sequence starts.
- Disqualifying questions in the call script. Whatever doesn't show up in a database has to be asked. And SDRs need to be measured on meeting quality, not just volume, so they have no incentive to ignore red flags.
That last point is where many teams stumble. If the only target is the number of meetings, the churn filter becomes an obstacle. Ideally, you track the rate of meetings that move to a proposal and, further down the line, retention of outbound-sourced accounts. At Nord, this kind of feedback from the client's sales team feeds into ICP and sequence adjustments throughout the engagement, precisely because a meeting with the wrong account wastes everyone's time.
The upside you don't expect
Excluding accounts feels like shrinking your market. In practice, it focuses your effort. With a smaller, better-fit list, sequences can be more personalized, messaging speaks to the real pain, and reply rates tend to go up. Your sales team also starts trusting the meetings they receive, which improves show rates and shortens the cycle.
There's a less obvious win, too: CS stops burning energy trying to save accounts that never should have signed, and starts investing in customers with real expansion potential.
Review it regularly
Your churn profile changes when your product changes, when your pricing changes, and when the market changes. A twice-a-year review, cross-referencing recent cancellations with account sources, keeps your criteria current. A company that was too small two years ago might become an ideal customer after you launch a new entry-level plan.
Next steps
Pull your list of cancellations from the past 12 months, separate the ones that left within six months, and sit down with CS for an hour to understand the real reason behind each one. Pick the two or three strongest patterns, turn them into list filters and qualification questions, and track over the next few weeks whether the meetings you generate move further down the funnel. It's a small adjustment that protects both your sales target and your recurring revenue.





