Section 01
The Business Situation at a Glance
Five numbers every decision-maker needs to understand the churn problem and the solution.
Churn Rate
0%
1,869 out of 7,043 customers left
1 in 4 customers
Potential Loss (4 months)
$0
1,869 churners × $64.76 ARPU × 4 months
Without intervention
Model AUC
0
Correctly ranks at-risk customers
Honest test
Net ROI with Model
$0
vs $12,903 doing nothing
↑ $14,520 improvement
Missed Churners (FNR)
0%
Customers flagged "safe" who left
57 customers · $3,933
Section 02
Why Do Customers Leave?
Three controllable factors explain the majority of churn. The good news: all three are actionable.
Factor #1 — Contract Type
Monthly subscribers are 15× more likely to leave than 2-year contract customers
42.7% of month-to-month customers churn vs just 2.8% on two-year plans. Converting a customer to an annual contract is the single most powerful retention lever available.
Factor #2 — How They Pay
Electronic check users churn at nearly 3× the rate of customers on automatic payment
45.3% churn on electronic check vs 15–17% on automatic methods. Customers who automate their payments are demonstrating commitment — and they stay.
Factor #3 — Service Bundle
Customers with no add-on services churn at nearly 3× the rate of bundled customers
Without online security: 41.8% churn. With it: 14.6%. Each service added creates switching cost — making it harder and more expensive for customers to leave.
The Danger Window
The first year is critical — nearly 1 in 2 customers in months 3–12 will churn
Churn peaks at 49.7% in the 91–365 day window — customers who haven't yet formed a strong habit or been locked into a longer contract. Onboarding programs during this window are highest-ROI.
Internet Service Type
Fiber Optic customers — the highest-value segment — churn at 2× the DSL rate
Fiber Optic: 41.9% churn · DSL: 19.0%. These are premium customers with higher monthly charges — their churn creates disproportionate revenue loss. They need differentiated retention.
Section 03
Which Customers Should Be Contacted?
The AI model scores every customer from 0 to 100% probability of leaving. Distribution across the 705-customer test cohort.
Risk Tier Breakdown — Test Cohort (705 customers)
125 customers flagged as High Risk — the model's priority call list
125
High Risk
Score ≥ 60% · Call immediately
73
Medium Risk
Score 30–60% · Monitor closely
507
Low Risk
Score < 30% · No action needed
The model flags 171 customers for intervention (24.3% of the cohort) — staying within the 25% call center capacity limit. Of those contacted, 76% are confirmed churners — 3 in every 4 retention calls are well-directed.
What a High-Risk Profile Looks Like
Maximum risk combination: Fiber Optic + month-to-month + electronic check
| Signal | Value | Risk Impact |
|---|---|---|
| Contract | Month-to-month | Very High |
| Internet | Fiber Optic | High |
| Payment | Electronic check | High |
| Tenure | 91–365 days | Very High |
| Add-on services | None | Elevated |
| Senior citizen | Yes | Elevated |
This profile reaches 53.7% churn probability — more than 1 in 2. These customers need proactive outreach within 48 hours.
Section 04
What Does This Mean for the Business?
Every retention call costs $20. Every successfully retained customer is worth $69 in recovered revenue. Here's the math.
Four Scenarios — Same Cohort of 705 Customers
The predictive model turns a $12,903 loss into a $1,617 gain
No action
$12,903
Do nothing · all churners lost
Default θ=0.50
$343
Standard threshold · 117 retained
Optimized model θ=0.41
$1,617
Optimized threshold · 130 retained
ROI lift vs default
+372%
Just by optimizing the decision rule
Key insight: The model does not just predict churn — it optimizes when to intervene. By lowering the decision threshold from 0.50 to 0.41, the system contacts more confirmed churners and generates $1,274 more in ROI from the same team.
Breakdown of the $1,617 Net ROI
Every call has a cost. Every retained customer has a value. Here's where the money goes.
The break-even point is straightforward: if at least 1 in 3 customers contacted is a real churner, the program pays for itself. The model achieves 3 in 4 — well above that threshold.
Section 05
Strategic Recommendations for the Retention Team
Five concrete actions, ranked by measurable impact on revenue retention.
01
Convert Monthly to Annual at Onboarding
Offer a discount or added service to new customers who commit to a 1 or 2-year contract. Month-to-month churn is 42.7% — two-year is 2.8%. This is the highest-leverage single action available.
15× churn reduction potential
02
Migrate Electronic Check Users to Auto-Pay
Incentivize customers paying by electronic check to switch to automatic bank transfer or credit card. This segment churns at 45.3% — 3× the rate of auto-pay customers.
3× churn reduction on migration
03
90-Day Onboarding Bundle Offer
Target customers in their first 365 days with a bundled offer (online security + tech support). Churn peaks at 49.7% in this window. Add-on services reduce switching cost and churn rate by up to 65%.
Critical 91–365d window
04
Senior Citizen Retention Program
1,142 senior customers churn at 41.7% vs 23.6% for non-seniors. A dedicated simplified service plan, dedicated support line, or proactive check-in calls would address this segment's specific needs.
1,142 customers at risk
05
Leverage the Model's Daily Risk List
The AI model generates a ranked list of at-risk customers. The retention team should work this list top-down every week. 76% precision means 3 of 4 calls reach real churners — far more efficient than random outreach.
76% call success rate