The $31.4M Question
Reconciling a $31.4M revenue shortfall to a zero residual loss bridge in four weeks. Then staying on to lead the product work that closes the leak.
- ROLE
- Product Manager
- SCOPE
- Loss bridge diagnostic, causal analysis, Voyager platform delivery
- SKILLS
- KPI definition, loss bridges, matched control analysis, SQL, Databricks, LLM assisted validation
- STATUS
- In delivery
Context
A global veterinary health network of more than 700 hospitals came out of a platform migration with revenue visibly eroding and no agreed answer for where or why. Averages hid the leak. Every top line metric looked survivable while the trend line said otherwise.
The problem
Revenue is down is a symptom, not a problem statement. The work was to reconcile a $31.4M shortfall into a loss bridge precise enough that every dollar had a mechanism, an owner, and a testable fix. The window: four weeks.
What I did
- Defined six canonical executive KPIs so leadership argued from one set of numbers instead of five versions of the truth.
- Ran eleven hypothesis driven investigation streams and pressure tested each until only the survivors remained.
- Designed matched control and same hospital pre and post causal analysis across 200 paired hospitals. It ruled out five popular hypotheses, including a write off metric that was overstated 50 to 90 times.
- Closed the bridge to zero residual: 85.9 percent of the loss was visit volume, 5.3 percent was pricing. The remediation roadmap moved from pricing to workflow fixes because the data said so.
Outcome
The diagnostic became a mandate. Since June 2026 I lead the product work on the Voyager platform closing more than $5M of that leakage at application level: validation blocking clinical codes on non visit orders, appointment status enforcement, and visit capture workflow fixes. A separate LLM validated reconciliation pipeline I scoped catches roughly 6 percent of missed billings, worth $16K to $20K a month.
What it taught me
Big numbers are hiding places. Keep dividing until a metric becomes someone's Monday morning. And a good enough diagnostic stops being a report. It becomes your job description.