Customer analytics 2021 to 2024

Scoring which customers were worth the call

Before the fraud work, the same ranking problem: a large business customer base, finite calling capacity, one score deciding the week.

Data Analyst, Customer Acquisition, AUTO1 Group

Impact

Replaced flat outreach across a large business customer base with a ranked call queue. Finite calling capacity went to the accounts carrying the most value.

What a value rank does to a week of calling relative value carried
  • Top-ranked accounts
    100
  • Next band
    58
  • Middle of the base
    31
  • Long tail
    12
Illustrative magnitudes, since the live score runs on AUTO1's customer data. What the work established is the ordering, and that the base was not evenly valuable. Flat outreach treats these four bands as though they were the same bar.
Show the data table
What a value rank does to a week of calling: ranked.
CategoryValue (relative value carried)Note
Top-ranked accounts100Worked first, because the rank put them first rather than because they called in.
Next band58
Middle of the base31
Long tail12
Context

AUTO1's commercial team worked a large base of business customers with finite outreach capacity. With no ranking over it, effort spread evenly across customers who were not evenly valuable. The most valuable accounts got the same share of the week as the least.

What I did
  • I built the value score that identified the high-value business customers worth proactive attention.
  • I turned it into a ranked outreach queue the commercial team worked from the top down.
  • That sent finite calling capacity to the accounts carrying the most value.
  • I kept it separate from the churn-risk score, because who is worth keeping and who is leaving are different questions.
Outcome

Outreach ran off a queue instead of spreading evenly, so the rank decided the week rather than whoever happened to call in.

Without this work

Calling capacity keeps being spent evenly across a base that was never evenly valuable, and the largest accounts get the same week as the smallest.

The full story

Fraud came later. The question did not change. A commercial team with finite outreach capacity and a customer base that is not evenly valuable faces the same problem as a fraud analyst holding a review queue: the population is large, attention is scarce, and spending it evenly is the one strategy guaranteed to be wrong. Unranked, the most valuable accounts got the same share of the week as the least, and the accounts quietly on their way out got none of it.

So I built the value score as its own number rather than blending it with churn risk, because who is worth keeping and who is leaving are different questions, and answering them in one figure hides which of the two is driving a call. The churn side became its own piece of work, and its own case.

What the score changed was mundane and it was the whole point. The team stopped deciding the week by whoever happened to ring in, and started at the top of a list.

This role was the training ground for everything after it. Population-level scoring, ranking under a hard capacity constraint, and the discipline of quantifying a decision before asking anyone to make it are the same habits the fraud work runs on. What changed later was the cost of being wrong. A missed call became a shut-down account. The structure of the problem did not move.

  • Customer scoring
  • B2B
  • Prioritisation
  • Revenue analytics