43.4% of pharmacy spend was detectable anomaly. Observed, not modeled.

The 43.4% number, unpacked

43.4% is the aggregate anomaly rate Inspector AI observed on a single production book of roughly 50,000 Latin American pharmacy subscribers over one year. It is not a projection. It is the share of pharmacy spend where the detection layer identified anomalies the insurer had not flagged. The number decomposes into five WAFL categories, each observed independently. A different book will produce different percentages — these are the ones we saw on this one.

• Waste and over-utilization: 20.3% of pharmacy spend
• Abuse — generic substitution opportunity: 10.8% of pharmacy spend
• Lack of Knowledge — clinical mismatch: 7.3% of pharmacy spend
• Fraud — behavioral risk, cloned scripts: 4.6% of pharmacy spend
• Financial anomaly: 0.4% of pharmacy spend

The WAFL categories and what they cost

WAFL — Waste, Abuse, Fraud, and Lack of Knowledge — is the framework behind every number on this page. Each category has a different detection strategy because each represents a different way money leaks out of a pharmacy benefit program.

• Waste: overuse and over-utilization. Not fraud, not malicious — just more drug than the patient needs, for longer than the patient needs it.
• Abuse: legal but exploitative. Generic substitution arbitrage, branded dispensing when a clinically equivalent generic exists.
• Fraud: intentional deception. Cloned prescriptions, same-molecule re-authorization rings, behavioral fraud patterns.
• Lack of Knowledge: clinically wrong or rule-violating approvals by doctors and reviewers who did not know the guideline. No malicious intent. Just uninformed.

What we observed on a 50,000-subscriber book

These are the numbers Inspector AI measured on one real Latin American pharmacy benefit program of roughly 50,000 subscribers, over one production year. They are observations on a single book, not projections for yours. The median per-dispensation savings on this book are shown alongside each total — those medians will shift on a different book with a different drug mix, a different regulatory environment, and a different prescribing culture.

• Generic substitution opportunity: $4.53 million observed, at a median of $16 per dispensation on this book
• Same-molecule re-authorization: $244,000 observed, at a median of $13 per dispensation
• Early refill: $176,000 observed, at a median of $41 per dispensation
• Clinical mismatch: $126,000 observed, at a median of $22 per dispensation
• Cumulative dose: $13,000 observed, at a median of $14 per dispensation
• Cloned prescriptions: $8,000 observed, at a median of $26 per dispensation
• Total detectable loss observed on the book: $5.1 million in one year

Why observed beats modeled

Most fraud detection vendors sell on projected ROI. A projection is a claim about the future you cannot verify until it arrives. Observed outcomes on real data are harder to fake because they already exist — you can audit them. Inspector AI publishes its observations because the alternative is asking you to trust a projection, and any vendor unwilling to share observed outcomes on real claims is asking for trust they have not earned. The three-week proof of concept is the same measurement run on your book. Your numbers will not match ours — they will be your own. That is the point.