Doctor Penetration
- SQL
- Power BI
- →46,552 doctors classified across 15 APAC markets
- →Case-level submission method resolved entirely in SQL
- →Three behaviour-based segments: PVS, competitor, iTero
Context
The business wanted a clear picture of how Invisalign cases were being submitted across APAC and what tools doctors used to do it. Every case can be traced from submission through to manufacturing or cancellation, and the tool behind each one matters: traditional PVS impressions, a competitor's scanner, or an iTero scanner. Knowing which doctors sat in which camp would show sales and marketing where conversion effort was worth spending.
The data was there, the logic was not
Raw submission records existed, but nothing turned them into a doctor-level picture. I built the classification entirely in SQL: working through large submission datasets to resolve the submission method for each case, then rolling cases up into a behaviour profile for each doctor.
Segmentation rules
From there I defined rules to label each doctor, and the hierarchy was behaviour-based rather than a simple majority. A doctor with ten competitor-scanner submissions is a competitor doctor; a doctor with a mixed history but more than three iTero submissions counts as an iTero doctor. The thresholds were set so a doctor who had genuinely started using iTero was not buried by a long tail of older competitor activity.
Result
The finished report classified 46,552 doctors and was surfaced in Power BI with regional breakdowns and filtering. Marketing used it to build target lists for scanner-conversion campaigns; sales used it to tailor outreach depending on whether a doctor was locked into a competitor ecosystem or already drifting toward iTero. It became a standing input to commercial strategy.
What I'd improve
The thresholds are hand-set. Tuning them against actual conversion outcomes, or replacing the rule with a recency-weighted score, would make the segments harder to argue with.