12 early warning signs a customer is about to churn
Concrete signals you can check this week in your existing tools, before the cancellation email arrives.
THE SHORT ANSWER
The 12 most reliable early warning signs of B2B SaaS churn are: declining daily active users, shrinking session lengths, delayed support ticket responses, missed onboarding milestones, failed payment attempts, reduced feature usage, fewer login frequency, skipped team invitations, ignored renewal reminders, reduced admin engagement, shortened contract terms, and fading communication patterns. You can check each of these in your existing tools this week, without installing anything new.
Why these 12 signals matter more than NPS
Net Promoter Score tells you how a customer felt six months ago. The signals on this list tell you how they are behaving right now.
Behavioral signals are consistently more predictive than attitudinal ones. A customer who gives you a 9 on your last NPS survey can still churn if their usage dropped 40 percent last month. A customer who rates you a 6 but uses your product daily is not going anywhere.
The 12 signs below are organized by where you can find them: product analytics, support tools, billing systems, and your CRM. Check each one this week. You will likely find two or three accounts showing multiple signals simultaneously. Those are your priority save calls.
Product analytics signals
These signals require access to your product analytics tool (Amplitude, Mixpanel, Heap, or your own database).
Sign 1: Daily active user decline
When the number of daily active users on an account drops by more than 30 percent compared to their 30-day average, that account is showing the earliest and most reliable churn signal.
What it looks like: An account that averaged 12 daily active users last month is now averaging 8 or fewer.
Why it predicts churn: Usage is the strongest leading indicator of retention. Totango's analysis of B2B SaaS churn patterns found that declining usage appears 60 to 90 days before cancellation on average.
Check this week: Pull your usage report for the last 30 days. Sort by percent change from the prior month. Any account down more than 30 percent belongs on your watch list.
Sign 2: Shrinking session duration
When users spend less time per session than they used to, they are getting less value from each interaction.
What it looks like: Average session length drops from 12 minutes to 7 minutes over a 30-day period.
Why it predicts churn: Shorter sessions mean users are doing less with your product. They may have found workarounds, or they may be using competing tools for the same workflow.
Check this week: Compare average session duration this month to the prior month for each account. Flag any account with a 35 percent or greater decline.
Sign 3: Feature adoption stall
When an account stops exploring new features or retreats to a narrower set of functionality, they are consolidating their usage before leaving.
What it looks like: An account that previously used 8 of your 15 features is now using only 3. Or an account that adopted a new feature in month one has stopped using it entirely.
Why it predicts churn: Feature breadth correlates strongly with switching cost. The more features an account uses, the harder it is to leave. When they narrow their usage, they are reducing their switching cost intentionally.
Check this week: Look at feature usage by account. Any account that has dropped below 50 percent of their peak feature count is signaling.
Sign 4: Fewer logins per user
When individual users on an account log in less frequently, the account is decaying from the inside.
What it looks like: Users who logged in daily are now logging in twice a week. Or users who logged in weekly have stopped entirely.
Why it predicts churn: Login frequency is a micro-habit signal. When the habit breaks, the dependency on your product breaks with it.
Check this week: Compare login frequency per user over the last 30 days to the prior 30 days. Any user with a 50 percent or greater decline in login frequency is worth flagging.
Sign 5: Fewer team invitations
When an account stops inviting new team members, they have decided your product is not worth expanding.
What it looks like: An account that previously invited 3 to 5 new users per quarter has invited zero in the last 90 days.
Why it predicts churn: Team expansion is the strongest signal of internal adoption. When it stops, the champion has lost confidence or moved on.
Check this week: Look at user growth by account over the last 90 days. Flat or declining user counts are a signal, especially for accounts in their first year.
3xhigher churn risk when team invitations stall
Support desk signals
These signals require access to your support tool (Zendesk, Intercom, Freshdesk, or similar).
Sign 6: Delayed ticket responses on your side
When your support team takes longer to respond to an account's tickets, that account is building frustration.
What it looks like: Average first response time for an account goes from 4 hours to 24 hours or more.
Why it predicts churn: Slow support does not just frustrate the person who filed the ticket. In B2B, it signals to the entire account that they are not a priority. Zendesk's customer experience research found that response time directly correlates with retention across B2B accounts.
Check this week: Filter your support tool by account. Sort by average first response time. Any account with response times above your SLA belongs on your watch list.
Sign 7: Unresolved tickets piling up
When an account has multiple open tickets with no resolution, they are accumulating evidence that your product does not work for their use case.
What it looks like: An account with 4 or more open tickets older than 7 days.
Why it predicts churn: Each unresolved ticket is a reminder that your product has failed them. The more tickets pile up, the stronger the narrative becomes that your product is not the right fit.
Check this week: Pull a report of open tickets by account. Any account with 4 or more tickets older than 7 days needs immediate attention.
Sign 8: Shift from questions to complaints
When an account's support tickets change from "how do I" to "this does not work," they have moved from learning to blaming.
What it looks like: Tickets that used to ask about features now report bugs or express frustration with basic functionality.
Why it predicts churn: Question-type tickets indicate engagement. Complaint-type tickets indicate loss of confidence. The shift happens gradually, and by the time it is obvious, the account may already be evaluating alternatives.
Check this week: Read the last 10 tickets from each at-risk account. Categorize them as questions or complaints. If the ratio has shifted toward complaints, that is a signal.
Billing and payment signals
These signals require access to your billing system (Stripe, Chargebee, Recurly, or similar).
Sign 9: Failed payment attempts
When a payment fails and the customer does not immediately update their card, they are either in financial distress or they do not care enough to fix it.
What it looks like: A failed payment followed by no action for 7 days or more.
Why it predicts churn: Stripe's research on reducing involuntary churn found that accounts with unresolved failed payments churn at 2 to 3 times the rate of accounts with clean payment history. The payment failure is both a technical issue and a commitment signal.
Check this week: Pull a list of failed payments from the last 30 days. Any account with an unresolved failure is at higher risk.
Sign 10: Downgrade requests or plan changes
When an account requests a downgrade, they are explicitly telling you they want to pay less for your product.
What it looks like: A request to move from annual to monthly billing, from a higher tier to a lower tier, or to reduce the number of seats.
Why it predicts churn: Downgrades are partial churn. The account is reducing their commitment before eliminating it entirely. A ProfitWell analysis found that 30 to 50 percent of downgrades precede a full cancellation within 12 months.
Check this week: Look at any downgrade requests from the last 60 days. Each one is a save opportunity.
Sign 11: Reduced contract length
When an account switches from annual to monthly billing, they are reducing their switching cost and giving themselves an exit door.
What it looks like: An account that previously renewed annually is now requesting month-to-month billing.
Why it predicts churn: Annual contracts are a commitment signal. Monthly contracts are a flexibility signal. When an account moves from annual to monthly, they are explicitly preserving the option to leave.
Check this week: Review your upcoming renewals. Any account requesting shorter contract terms is signaling reduced commitment.
CRM and communication signals
These signals require access to your CRM (Salesforce, HubSpot, or similar) and your email or Slack history.
Sign 12: Silence from the champion
When your primary contact stops responding to emails, skips QBRs, and goes dark on Slack, the most likely explanation is that they are no longer your advocate inside the account.
What it looks like: Three or more unanswered outreach attempts over 30 days. Skipped quarterly business review. No response to renewal reminders.
Why it predicts churn: In B2B, the champion is the single biggest predictor of renewal. When they disengage, there is no one inside the account advocating for your product. A TSIA research report found that champion turnover or disengagement is the top predictor of B2B churn, ahead of product issues and pricing.
Check this week: Look at your last 10 outreach attempts to each account's primary contact. Any account with 3 or more unanswered messages needs a different approach, not the same template sent again.
3xhigher churn risk when champion goes silent
What to do with this list
Reading this list and doing nothing is the same as not reading it. Here is how to turn these signals into action:
Step 1: Pick your top 10 accounts. Choose the 10 accounts showing the most signals across the categories above. Prioritize by ARR.
Step 2: Check the reason. For each account, identify which signals are firing. This tells you what type of intervention they need.
Step 3: Make the save call. Reach out with a specific reason, not a generic check-in. "I noticed your team's usage dropped 30 percent last month. I want to make sure you are getting value from the onboarding workflow we set up" is better than "Just wanted to touch base."
Step 4: Track the outcome. Did usage recover? Did the signals improve? The feedback loop is what makes your prediction better over time.
This is not a one-time exercise. Build these checks into your weekly rhythm. The accounts that show up on next week's list are the ones you save this week.
Related resources
RELATED
01 ·Churn prediction for B2B SaaS: the complete guide
How to predict which customers will leave, score risk accurately, and run a save workflow without a data science team.