The Setup
TaskFlow, a B2B project management SaaS, had a churn problem. 4.2% monthly. High for a mature product.
The exit surveys weren't helping. Every response was a variation of "switched to competitor" or "not a good fit." Which tells you they left. Not why.
So the team conducted 15 exit interviews with churned customers and pulled their usage data — feature adoption, tenure, support tickets, plan tier — for the 90 days before cancellation.
What they found surprised everyone: the same churn rate was hiding two completely opposite problems.
The Data
What we had:
- 15 exit interview transcripts (DesignWorks, TechOps, MarketPros, AgencyOne, and others)
churned_customers.csv: usage metrics including feature adoption %, months active, plan tier, support ticket count, MRR
| Cohort | Feature Adoption | Avg Tenure |
|---|---|---|
| "Too Complex" | 13% | 5.5 months |
| "Missing Features" | 83% | 35 months |
The Investigation
Qualitative Coding
Five themes emerged from the exit interviews:
- Too Complex — overwhelmed by features, never found the core workflow
- Missing Features — outgrew the product, building workarounds at scale
- Poor Mobile — field teams blocked by desktop-centric design
- Price Sensitivity — usage didn't justify cost after a price increase
- Support Issues — slow response times, unresolved tickets
DesignWorks' interview (8-person design agency, $49/mo, 6 months):
"TaskFlow just got too complicated for us. All these features we didn't understand."
"There's a simplified view? No one told us that."
DesignWorks churned because they were overwhelmed by a product that had evolved beyond their use case. The "simplified view" quote was the most damning — a feature existed that would have solved their problem, and they never discovered it. That's an onboarding failure, not a product failure.
TechOps' interview (150-person engineering org, $890/mo, 38 months):
"We outgrew it. We need custom fields, API access, proper automation."
"Three years of roadmap promises and still no custom fields."
TechOps was the opposite of DesignWorks in every measurable way. Power users for three years. 83% feature adoption. $890/mo. They didn't find TaskFlow too complicated — they found it too limited. And they had waited three years for the roadmap to catch up.
MarketPros revealed a third distinct profile:
"The mobile app is basically unusable."
A field-first team whose primary use case was categorically underserved.
AgencyOne added the price-sensitivity pattern:
"You raised prices 40%. We were only using maybe 30% of the features."
Low adoption combined with a price increase. The value equation broke.
The Joint Display: Where It Came Together
Split View: Two Profiles, Side by Side
The Split View joint display made the contrast visible in a single screen.
Filtering to "Too Complex":
Left panel — customer quotes:
- DesignWorks: "All these features we didn't understand"
- FreshBake: "We only needed basic task lists"
- SmallBiz: "There's a simplified view? No one told us that"
Right panel — usage metrics:
- Feature adoption: Mean 13% (Min 8%, Max 19%)
- Average tenure: 5.5 months
- Average support tickets: 1.2
- Plan tier: Mode Starter
Switching filter to "Missing Features":
Left panel — customer quotes:
- TechOps: "We outgrew it. We need custom fields, API access."
- DataForge: "Three years of roadmap and still no custom fields."
- ScaleUp: "We're building workarounds for everything at this point."
Right panel — usage metrics:
- Feature adoption: Mean 83% (Min 76%, Max 91%)
- Average tenure: 35 months
- Average support tickets: 47
- Plan tier: Mode Enterprise
47 support tickets per account — not because the product was broken, but because these customers were constantly pushing against its limits. DataForge waited three years for custom fields. That's loyalty that was burned through.
The Integration Matrix: Confirming the Structure
| Theme | features_adopted_pct | months_active | mrr |
|---|---|---|---|
| Too Complex | Strong negative | Strong negative | Strong negative |
| Missing Features | Strong positive | Strong positive | Strong positive |
| Price Sensitivity | Weak negative | Moderate negative | Moderate negative |
| Poor Mobile | Neutral | Moderate negative | Neutral |
"Too Complex" and "Missing Features" are not variations of the same problem. They're inverse problems with inverse solutions.
The Recommendation
Three interventions, addressing three distinct churn profiles:
1. For "Too Complex" churners: proactive onboarding intervention These customers never discovered the product's core value. The fix is not simplifying the product — it's detecting the failure pattern early. An SMB account at 3 months with <20% feature adoption is at risk. Trigger a proactive check-in before they give up.
DesignWorks didn't know the simplified view existed. That's a discoverability failure with a measurable early-warning signal.
2. For "Missing Features" churners: customer success escalation Enterprise accounts using 80%+ of available features and raising the same requests repeatedly are on a countdown. Assign dedicated customer success to these accounts before renewal conversations begin. TechOps gave three years of patience — the product team owed three years of honesty in return.
3. Track churn by profile, not by aggregate rate A 4.2% monthly churn rate is a number. "Too Complex" churn and "Missing Features" churn are two different business problems:
- "Too Complex" segment: low MRR, short tenure — high volume, low lifetime value loss
- "Missing Features" segment: high MRR, long tenure — lower volume, catastrophic lifetime value loss
Same symptom. Opposite treatments. Reporting them as one metric obscures both.
The Outcome
Before the analysis, the product team was treating all churn as one problem. Exit surveys weren't granular enough to reveal the split. The aggregate churn rate looked like one thing to fix.
After the integrated analysis: two profiles, two intervention strategies, two success metrics.
The "Too Complex" fix is an onboarding product change — detectable and testable within weeks. The "Missing Features" fix is a roadmap commitment — slower, but the ARR impact is orders of magnitude larger.
The qualitative showed why customers left — in their own words, with the specific frustrations that drove the decision. The quantitative showed the usage pattern that preceded the churn — so both profiles can be detected before the cancellation email arrives.
That's the difference between reactive churn analysis and predictive intervention.