Case study
A shrinking customer base, reversed in about ten weeks.
A founder-led B2B software company of around fifty people was losing customers faster than it could win them. It had outgrown the instinct and memory that built it, so we put measurement where there was none, and the numbers moved.
What we did
Verify, measure, then fix.
The same method, in the same order, behind every engagement.
Verify what's real
The company believed it served roughly ten thousand customers. Cross-referencing billing against the CRM, we found more than a thousand ghost records, test accounts, duplicates, and cancellations that were never closed, and set the real paying base at less than half the assumed figure. A reconciliation across four independent sources tied out to zero variance and surfaced more than 450 accounts that were billing but missing from the CRM. The same pass showed cancellations outrunning new sales by more than three to one, arriving through three separate inboxes, with roughly 44% sitting in one uncategorised bucket, so no one could say why customers were leaving.
Put measurement in place
Before changing anything, we built the instruments the company lacked. A 21-category system captured the real reason behind every cancellation, with the fields enforced so the data stayed clean. We analysed more than 8,800 support tickets to set the first real productivity and response baselines, and added service-level and satisfaction tracking where feedback had been running under 2%. A supplier-by-supplier cost model, cross-referenced across five sources, flagged about $31,000 a year of spend that was not paying its way inside a vendor base near $146,000.
Fix what's costing the most
With the picture clear, we made the few changes the numbers pointed to. The largest was a structured retention-call routine, tested personally on the first calls, which saved half of the customers who called to cancel against a prior rate under 10%. Categorised cancellations made the biggest recoverable driver visible and workable, and the supplier model became a repricing list.
The results
About ten weeks, measured.
Net customer movement
−68 / mo→+25 / mo
Cancellations to new sales
3.2 to 1→1.35 to 1
Effective annual churn
24.4%→16.9%
Retention-call save rate
under 10%→50%
Open support backlog
571→375
Measured over about ten weeks. Figures rounded; client details withheld for confidentiality.
What the company kept, built to run without us:
Sound like your business?
A Business Check-Up runs this on your numbers. A short, confidential call, no pitch and no obligation.