How Rx Attribution Works: Matches, Windows, and Comparison Groups
A plain explanation of how pharma Rx attribution links ad exposure to prescriptions, where the data can fail, and what questions to ask a measurement partner.
The short answer
Rx attribution connects people or prescribers who saw an ad with prescriptions filled or written afterward. Brands use it to judge partners and campaigns. Many decision makers see the output, a number of attributed prescriptions, without understanding how it was produced. Knowing the steps makes it easier to spot where results can go wrong.
This is a general explanation. Each measurement vendor has its own methods, which you should review directly.
Step 1: Record exposure
When an ad is served, the activation partner records an identifier for the device or person who received it. For HCP campaigns, this is usually linked to an NPI. For DTC campaigns, it is often a device or household identifier.
Where it fails: not every impression can be tied to an identifier the measurement partner can use.
Step 2: Create a privacy-safe link
For DTC measurement, personal identifiers are usually converted into de-identified tokens using a process that allows records to be matched without exposing the underlying identity. This is typically done by a specialized partner under a documented privacy methodology. Your privacy team should review how it works.
Where it fails: tokens from exposure and tokens from prescription data must match. Differences in how identity is resolved can lose records. See why Rx match rates drop.
Step 3: Match to prescription data
The measurement partner links exposed tokens or NPIs to prescription data sources, usually licensed pharmacy or claims data. Coverage varies. No source captures every prescription.
Where it fails: gaps in data coverage mean some prescriptions are never observed.
Step 4: Apply a window
Only prescriptions that occur within a set period after exposure are counted. The window reflects assumptions about how long media takes to influence behavior. See choosing an attribution window.
Where it fails: too short a window misses real effects; too long a window counts prescriptions unrelated to media.
Step 5: Compare to a baseline
A good study compares exposed people or prescribers with a similar unexposed group. The difference estimates lift. Without a comparison group, the result is attribution only.
Where it fails: if the comparison group differs in ways that affect prescribing, the lift estimate is biased. Randomized holdouts reduce this risk.
Questions to ask your measurement partner
| Question | Why it matters |
|---|---|
| What share of exposures could be linked? | Shows how much data was lost |
| What prescription data sources are included? | Shows coverage gaps |
| How was the comparison group built? | Determines whether lift is credible |
| What window was used, and why? | Affects what is counted |
| What is the uncertainty range? | Shows decision risk |
Reading the report
Look for the match funnel first, then the outcome definition, then the design. A large attributed number with a thin match funnel and no comparison group says less than a small lift estimate with a solid design. The measurement design guide explains which designs support which decisions.
Practical takeaway
Ask every measurement partner for a one-page summary covering the five steps above: how exposure was captured, how records were linked, what data sources were used, what window applied, and how the comparison group was built.
Sources
- HHS, Use of Online Tracking Technologies by HIPAA Covered Entities
- FTC, Health Breach Notification Rule: The Basics for Business
- Media Rating Council, Standards and Guidelines
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.