Why an Rx Match Rate Drops and How to Diagnose It
Diagnose Rx match-rate declines across eligibility, identifiers, identity resolution, consent, platform delivery, claims coverage, and timing.
The short answer
When an Rx match rate drops, compare each stage with the last known-good period: submitted records, valid identifiers, eligible records, identity-resolved records, addressable exposures, claims-covered records, and reportable outcomes. A single blended percentage hides where loss occurred.
Reconstruct the funnel
| Stage | Common failure | Evidence |
|---|---|---|
| Intake | File format, hashing, schema, duplicate change | Row counts and rejection codes |
| Eligibility | Geography or suppression rule changed | Before/after exclusions |
| Identity | Graph refresh, consent, identifier loss | Match by identifier type and cohort |
| Delivery | Channel mix or browser/app shift | Exposure distribution |
| Outcome | Claims coverage, lag, outcome definition | Coverage and maturity tables |
| Reporting | Threshold or privacy rule | Suppressed-cell log |
Define numerator, denominator, unit, and date for every rate. “Claims match” may mean eligible users linked to claims, exposed users linked to claims, or conversions linked after exposure.
Segment the change
Compare new versus returning records, channel, device, publisher, geography, audience source, HCP specialty, and campaign cohort. A broad decline suggests pipeline or policy change; a concentrated decline suggests composition or delivery.
Check timing before performance
Claims feeds and adjudication can mature after media reporting. Build maturity curves showing how outcomes accumulate by days since exposure. Do not interpret an early partial window as a permanent match decline.
Protect comparability
Version source files, identity logic, partner configuration, lookback, attribution window, outcome definition, and code. If two periods use different denominators, restate them before diagnosing performance.
Hypothetical diagnosis
Overall match falls from 62% to 48%. Intake and identity resolution remain stable, but CTV grows from 20% to 55% of exposures and uses a household unit with lower person-level claims linkage. The likely driver is mix and unit change, not sudden audience deterioration. Numbers are illustrative.
What not to do
Do not loosen eligibility solely to restore the rate. Do not swap identity partners before locating the break. Do not treat unmatched people as non-converters. Missing linkage can be systematically different.
Build a match-rate monitor
The best time to diagnose a match-rate drop is before anyone notices it in a quarterly report. A simple monitor makes that possible:
- Track each funnel stage weekly, not just the final rate.
- Set alert thresholds for each stage, for example a drop of more than five percentage points from the trailing four-week average.
- Log every change to source files, identity partners, targeting, channel mix, and measurement configuration, with dates.
- Check alerts against the change log first. Most drops line up with a known change.
Share the monitor with the measurement partner and activation partners. When everyone sees the same funnel, diagnosis becomes a joint exercise rather than a debate about whose data is right.
Also watch for match rates that rise suddenly. An unexpected increase can mean a partner changed its matching method, loosened criteria, or added probabilistic matches. That can make results look better while making them less reliable. See deterministic vs. probabilistic HCP identity for what to ask when a method changes.
Practical takeaway
Match-rate troubleshooting is data lineage work. The next step is a recurring funnel report with stage counts, definitions, version history, cohort cuts, and data-maturity status.
Sources
External guidance and platform documentation change. Links were current at publication; check them again before relying on them for a decision.
Editorial note. Analysis and frameworks are the author's own and do not represent Acxiom or any current or former employer, client, or named platform. Examples labeled hypothetical or illustrative are not results from real campaigns. Nothing here is legal, regulatory, or medical advice.
New pharma programmatic breakdowns, occasionally
One email when I publish something worth reading. Benchmarks, measurement teardowns, and case studies with the caveats attached. No cadence promises, no reselling your address.
Unsubscribe any time. See the privacy policy.
Working through this decision on a real plan?
I work on health and pharma data, identity, and activation, after five years running HCP and DTC programmatic agency-side. Happy to talk through how this applies to your situation.