How to Find the Usage Pattern That Predicts Which Customers Stay
Every product has a usage threshold that distinguishes between customers who stay and those who leave.
Cross it, and retention becomes nearly guaranteed. Fall below it, and churn is almost inevitable.
Most companies are unsure where this line lies. They’re tracking logins, page views, and engagement scores that don’t actually predict retention.
What I will cover in this article: How to identify the specific usage pattern that predicts long-term retention, how to find it in your data without a data science team, and how to use it to catch at-risk customers before they churn.
The Threshold Exists (And You Can Find It)
Retention data reveals a consistent pattern: there’s always a sharp dividing line, a usage threshold at which the probability of retention dramatically shifts.
Users above this line rarely churn. Users below it almost always do.
Documented examples:
Slack: Stewart Butterfield publicly shared that teams sending more than 2,000 messages showed 93% retention compared to under 30% for teams below that threshold.
Facebook (early growth period): Chamath Palihapitiya revealed their “7 friends in 10 days” activation metric - users who achieved it had exponentially higher retention.
Dropbox: Drew Houston explained that users who organize files into multiple folders and share at least one had a retention rate of 85% or higher, whereas single-folder users often churned.
Why This Changes Everything
Once you know your threshold, you can predict retention before it happens.
Look at any customer’s usage in their first 30 days, and you’ll know with high accuracy whether they’ll still be around in six months. You’re no longer reacting to churn, but you’re actually predicting it.
The change this causes:
Instead of asking “Why did they leave?” after they’re gone,
you’re asking “Who’s going to leave?” while you can still do something about it.
Instead of treating all customers the same, you focus your energy on those just below the threshold, the winnable saves who are close but need a push.
Instead of tracking vanity metrics that make you feel good (page views, time in app, etc.), you track the ONE behavior that actually predicts whether someone stays or leaves.
This is the difference between hoping customers stick around and knowing who will stay based on their actions in their first month.
Why Most Companies Track Useless Metrics
You’re probably measuring engagement wrong. Here’s what doesn’t predict retention:
❌ Total time in app (vanity metric—confused users spend lots of time too)
❌ Page views (clicking around ≠ value)
❌ Feature adoption rate (which features matter?)
❌ NPS Scores (lagging indicator, shows experience, not future behavior)
These metrics feel good, but they won’t tell you who’s about to churn.
How to Find Your Threshold
Step 1: Identify Your Core Value Actions
Not every action matters equally. You need to isolate the behaviors that directly correlate with the value your product delivers.
Questions to ask:
What action do customers take right before they say, “This is exactly what I needed”?
What do your most loyal customers do that churned customers don’t?
What behavior, when repeated, creates dependency on your product?
A few examples by product type:
Project management: Creating tasks = low signal. Assigning tasks to others + setting due dates = high signal
Analytics platform: Viewing dashboards = low signal. Creating custom reports + sharing them = high signal
CRM: Adding contacts = low signal. Logging activities + moving deals through the pipeline = high signal
If a customer stops doing this action for two weeks, would they even notice your product was gone? If no, it’s not a core value action.
Step 2: Analyze Your Retention Data
Pull data on customers from the past 6-12 months. Segment by retention outcome (retained vs. churned) and look for patterns.
Count how many times each user completed your core value actions during their first 30 days. Then compare the distributions between customers who stayed and those who left.
What you’re looking for: The inflection point where retention probability jumps significantly.
You’ll typically see something like this pattern:
Users with 0-3 actions: Low retention
Users with 4-6 actions: Moderate retention
Users with 7-10 actions: High retention
Users with 11+ actions: Very high retention
When you notice retention suddenly rise between two ranges (for example, from 40% to 75%), you’ve identified your threshold. That is your clear advantage.
Step 3: Define Your Activation Metric
Your threshold becomes your North Star. This is the specific, measurable behavior that predicts whether a customer will stay engaged.
Structure: “Users who [specific action] at least [X times] within [timeframe]”
Framework examples:
“Users who create multiple boards and invite collaborators in their first two weeks”
“Users who run several reports in their first month and export results”
“Users who complete multiple sessions and achieve a key milestone in their first three weeks”
Make it specific enough that anyone on your team can look at a user and tell you whether they’ve crossed the threshold or not.
The 4 Types of Thresholds (And How to Cross Them)
Different products have different types of thresholds. Understanding yours determines your intervention strategy.
Type 1: Frequency-Based
What it is: How often users return, not what they do.
Example: Communication platforms, project management apps. Slack’s message threshold isn’t about mastering features, but rather about building a daily habit.
How to drive it: Focus on creating routine triggers. Use notifications strategically, send digest emails at consistent times, and integrate with tools users already check daily.
Type 2: Depth-Based
What it is: It’s using multiple features together, not mastering one.
Example: Design tools, CRM systems, and analytics platforms. Users who only use one feature often fail to appreciate its value fully. The threshold is crossed when features work together.
How to drive it: Progressive onboarding that connects features. After they use Feature A, immediately demonstrate how Feature B enhances its power. Make the connection explicit.
Type 3: Network-Based
What it is: Value multiplies with team size or collaboration.
Example: Any tool with sharing or collaboration features. Individual users struggle; teams thrive.
How to drive it: Aggressively prompt team expansion. Make it clear that solo usage is suboptimal. Demonstrate precisely how adding teammates enhances the value of the product. Remove friction from invitations.
Type 4: Milestone-Based
What it is: There’s a specific achievement that proves value.
Example: Fitness apps (completing a workout), learning platforms (finishing a course), financial tools (hitting a savings goal).
How to drive it: Make the milestone visible and achievable in a short period. Break big goals into small wins. Don’t start with the end state; instead, start with something they can accomplish today that builds toward it.
Building Your Early Warning System
Once you know your threshold, you can spot at-risk users before they churn. You can do so by tracking these signals first.
🚨 Distance from threshold
How close are they to crossing?
How much time is left in their critical window (usually the first 30 days)?
🚨 Velocity changes
Is their activity declining week-over-week?
Slowing momentum is often more predictive than absolute numbers.
🚨 Stall patterns
Are they stuck? Started workflows but haven’t completed them?
Logged in but didn’t take action?
🚨 Incomplete value loops
Did they complete Steps 1 and 2 but never Step 3, where the payoff occurs?
Here is an intervention strategy based on risk that you can use
High proximity (70%+ to threshold): Automated encouragement. Show their progress and what’s next.
Medium proximity (40-70%): Personal outreach. Explain how similar users got unstuck and reached value faster.
Low proximity (below 40%) with limited time: Human intervention. Direct call, hands-on onboarding session, or personalized walkthrough.
Very low with deadline approaching: Last-chance offer. Remove all barriers and personally guide them through setup.
How Each Team Should Use This
Product: Redesign onboarding to push users toward the threshold. Every feature introduction, every tooltip, every email should be engineered to move them closer.
Customer Success: Segment by Distance from the threshold. Prioritize high-touch support for those who are close but stalling. These are winnable.
Marketing: Stop selling features. Sell the transformation that happens when users cross the threshold. Use proof from customers who’ve made it.
Leadership: Make threshold-crossing rate your primary health metric. It predicts future retention better than revenue or satisfaction scores. If more users cross the threshold this month than last month, you’ll see it in retention six months from now.
Not every user will cross your threshold. Some signed up for the wrong reasons, have the wrong use case, or aren’t willing to invest the effort.
That’s okay.
The goal isn’t to save everyone. The goal is to:
Know who CAN cross the threshold
Remove every obstacle preventing them from getting there
Intervene when they’re close but stalling
You can’t fix poor product-market fit just by hacking it. But if you’re providing real value and still experiencing high churn, this framework reveals exactly where users get stuck and how to help them move forward.