Churn Prevention Intelligence · Brief 02
High-value at-risk base · Chicago & Los Angeles
When the at-risk cohort is the profitable one
The one line you configure
How is the high-risk base moving this week and what is it worth — is risk concentrating in the cohorts we can least afford to lose, or the ones we can?
In a nutshell
It is the expensive kind. High-risk subscriber count across Chicago and Los Angeles rose 27% — and ARPU within that at-risk cohort rose 12% over the same period. Ordinarily those move in opposite directions. Together they mean the subscribers becoming risky are the ones we least want to lose: $4.1M of monthly revenue now sits in the high-risk band, up from $3.2M.
At a glance
WHAT MOVED · How the at-risk revenue built up · $3.2M → $4.1M
Three cohorts built the $0.9M increase and one recovered unaided — which is the difference between a blanket budget request and a targeted one.
WHERE IT IS · The at-risk base by city × plan × tenure band
| City × plan | Tenure | At-risk subs | Their ARPU | Revenue at risk |
|---|---|---|---|---|
| Chicago · Postpaid Gold | 3–6 yrs | 8,400 | $190 | $1.6M |
| Los Angeles · 5G Ultra | > 6 yrs | 4,900 | $224 | $1.1M |
| Chicago · Postpaid Gold | 1–3 yrs | 5,100 | $157 | $0.8M |
| Los Angeles · Postpaid Silver | 3–6 yrs | 4,100 | $98 | $0.4M |
| Philadelphia · mixed | < 1 yr | 3,600 | $56 | $0.2M |
Read three dimensions deep, the exposure is not where the subscriber count is: Philadelphia has 3,600 at-risk subscribers worth $0.2M, Los Angeles 5G Ultra has 4,900 worth $1.1M. Tenure is the discriminator — every expensive row is past three years.
WHY IT MOVED · Risk movement × ARPU × exposure, one plot
Bubble size is $ at risk, so the answer is the top-right corner: risk accelerating, ARPU high, exposure large. A count-ranked list would have put Philadelphia second — bottom-left, and the cheapest base in the region.
Why the weekly view matters
Daily score noise hides this. Rolled weekly, the divergence between a rising risk count and rising ARPU inside it is unmistakable — and it changes the retention budget question from how many to which.
So what
Retention budget should follow revenue, not headcount. The same spend aimed at the Chicago and Los Angeles high-ARPU, long-tenure cohort defends roughly $4.1M a month; spread evenly across the risk base it defends a fraction of that. Give retention the cohort ranked by revenue-at-risk rather than by score, and set the offer ceiling against each subscriber's own ARPU.
The points that matter
Value and risk are rising together
A 27% larger high-risk base whose ARPU is also up 12% — the opposite of normal churn behaviour.
+12%ARPU in cohortTwo cells hold two-thirds
Chicago Business and Los Angeles Premium account for $2.7M of the $4.1M exposed.
$2.7Mof $4.1MLong-tenure, not new joiners
Average tenure of 4.2 years means these are established relationships, which are both costlier to lose and cheaper to save.
4.2 yrsavg tenureQuestions it already answers
What is the offer ceiling that still pays back?
Sizes each cohort's revenue-at-risk against its own ARPU, so the retention offer has a defensible upper bound per subscriber.
What can this not tell us?
Not which offer they will accept — that needs a test. It names the cohort worth testing on and the budget it justifies.