Scoring which customers were worth the call
Scoring systems that told AUTO1's commercial team which B2B customers were worth pursuing, which were about to leave, and where retention money would actually pay back.
Data Analyst, Customer Acquisition
Turned an unranked B2B customer base into a prioritized outreach queue with value scoring, churn-risk flags, and a quantified revenue case for the highest-impact retention plays.
AUTO1's commercial team worked a large B2B customer base with finite outreach capacity. Without a ranking, effort spread evenly across customers who were not evenly valuable, and churn was discovered when it happened rather than before.
- I built the scoring systems that ranked the base two ways.
- A value score identified high-value B2B customers worth proactive attention, and a risk score flagged accounts whose behaviour signalled churn.
- Alongside the scores, I analysed customer drop-off patterns to locate the highest-impact retention opportunities and quantified the revenue upside of acting on them, so the case for retention work was a number rather than a feeling.
The commercial team could prioritize outreach instead of spreading it evenly. Value scores decided who to call first, churn scores decided who to save, and the quantified drop-off analysis decided where retention investment would actually pay back.
Outreach capacity keeps spreading evenly across an unevenly valuable base, and churn keeps being discovered in the revenue numbers after the customers are already gone.
Read the full story
Before fraud, the question I worked on was gentler but structurally identical: in a large population, which accounts deserve attention first?
AUTO1’s commercial team had finite outreach capacity and a B2B customer base that was not evenly valuable. I built the two scores that ranked it. A value score identified the high-value customers worth proactive attention, and a churn-risk score flagged the accounts whose behaviour said they were about to leave. Together they turned a flat customer list into a queue: who to call first, who to save this week.
The third piece made the case for acting. I analysed drop-off patterns to find where customers were actually being lost, then quantified the revenue upside of the highest-impact retention opportunities. Retention investment stopped being argued from anecdote and had a number attached instead.
Looking back, this role was the training ground for the fraud work that followed: population-level scoring, ranking under capacity constraints, and the discipline of quantifying a decision before asking anyone to make it.