Which accounts were leaving, and which losses were worth fixing
Two scores for two questions: which business customers were at risk of leaving, and which points of customer drop-off carried enough revenue to be worth working.
Data Analyst, Customer Acquisition, AUTO1 Group
Churn risk became a ranked outreach priority for the commercial team. Every drop-off point carried a quantified revenue upside, which is what identified the highest-impact retention opportunities.
- Top drop-off point by revenue 100
- Second by revenue 71
- Third by revenue 44
- Fourth by revenue 26
Show the data table
| Category | Value (relative revenue upside) | Note |
|---|---|---|
| Top drop-off point by revenue | 100 | The highest-impact retention opportunity, which is what the revenue figure was there to identify. |
| Second by revenue | 71 | |
| Third by revenue | 44 | |
| Fourth by revenue | 26 |
AUTO1's commercial team worked a large base of business customers with finite outreach capacity. Retention needs two things a customer list does not carry on its own: which accounts are at risk of leaving, and which points of customer drop-off have enough revenue behind them to be worth acting on.
- I built the churn-risk score that identified which business customers were at risk of leaving.
- I kept it separate from the value score, because who is worth keeping and who is leaving are different questions.
- I analysed customer drop-off patterns to find where in the customer relationship the base was being lost.
- I quantified the revenue upside sitting behind those drop-off points, which is what made the retention opportunities rankable by size rather than only nameable.
- The commercial team used both scores to prioritise its outreach.
Outreach was prioritised against churn risk rather than spread across the base. Each retention opportunity carried a revenue figure, so the highest-impact ones could be identified as the highest-impact ones.
Retention effort spreads evenly across a base where the losses are not evenly sized, and the largest recoverable losses get the same attention as the smallest.
A business customer never resigns. The account just stops buying, which makes churn a measurement problem before it is a retention problem.
Two questions sit inside it, and I answered them with two separate numbers on purpose. The value score said which accounts were worth keeping. The churn-risk score said which ones were at risk of leaving. Blended into one figure they hide which of the two is driving a call, and the commercial team working the outreach needed to know which it was.
A score names who. It does not locate where. So the second piece was reading customer drop-off as a pattern rather than as a single event, and asking at which points in the relationship the base was actually being lost. That turns churn from one number into a set of specific places to look.
The last piece is the one that made the set usable. Each of those drop-off points had revenue sitting behind it, so I quantified the upside at each one. Without that figure, a list of places the base drains is a list of everything, all of it plausible and none of it ordered. With it, the highest-impact retention opportunities are simply the ones carrying the largest number, which is something a commercial team can prioritise against.
Ranking a large population under a hard capacity constraint, and putting a figure on a decision before asking anyone to make it, are the habits the fraud work later ran on.