How to Attribute Prescription Lift to Programmatic HCP Media

The question usually comes from finance or a brand lead who has seen a deck claiming thousands of "attributed" scripts and wants to know if any of them are real. They are right to ask. Most HCP attribution reports are honest about their math but vague about their comparison group, and the comparison group is what turns a count into a lift.

Below is the process I would follow for a single programmatic HCP tactic. The concepts are covered in more depth in how Rx attribution works. This article is the operational version.

Step 1: Build a clean exposure file

The exposure file is the record of which prescribers saw which ads, and when. Everything downstream depends on it.

• Unit: one row per NPI per impression, or per NPI per day with an impression count.
• Fields: NPI, timestamp, tactic or line item, creative ID, publisher or deal ID.
• Source: the DSP or HCP data partner that resolved the impression to an NPI. Note whether that resolution was deterministic (logged-in professional site, verified email) or probabilistic. See deterministic vs. probabilistic HCP identity.

What usually goes wrong: the exposure file only covers some tactics, so exposed prescribers in the "control" are actually seeing your ads through another partner. Collect exposure from every HCP tactic, even ones you are not measuring, so you can exclude cross-exposed NPIs from the control.

Step 2: Match exposed NPIs to prescription data

The measurement partner joins exposure NPIs to prescriber-level Rx data, usually claims. Ask for the match rate and its denominator. A match rate of 85 percent against your target list means something different from 85 percent against NPIs that received at least one impression. The Rx match rate troubleshooting guide covers the common causes when this number drops.

Also confirm how the data handles prescribers who write through specialty pharmacy or mail order. If your brand runs through specialty channels, a retail-heavy dataset may undercount outcomes for exactly the prescribers you care about.

Step 3: Define test and control groups

This is the step that decides whether you get lift or just attribution.

For holdout setup basics, see what is a holdout test. For the advanced version with sizing and contamination rules, see how to design a holdout test for pharma media.

Step 4: Set the pre-period and the attribution window

The pre-period is the stretch before launch used to confirm that test and control behaved the same. Twelve to twenty-six weeks is common for established brands. For a launch brand there may be no pre-period on the brand itself, so use the competitive class or the condition.

The attribution window is how long after exposure you count outcomes. Too short and you miss delayed effects. Too long and you count scripts driven by everything else that happened. Pick it before the readout, not after. The tradeoffs are in how to choose an Rx attribution window.

Step 5: Calculate lift (a worked example)

All numbers here are hypothetical and chosen to make the math easy to follow.

A brand runs a 12-week programmatic HCP tactic. 10,000 NPIs are exposed. A matched control of 10,000 unexposed NPIs is built on specialty, region, decile, and pre-period writing. The outcome is new-to-brand prescriptions (NBRx) per prescriber over 12 weeks.

1. Incremental NBRx per exposed HCP: 0.15 minus 0.07 equals 0.08.
2. Total incremental NBRx: 0.08 times 10,000 equals 800.
3. Expected exposed post-period without media: 0.80 plus 0.07 equals 0.87 per HCP.
4. Percent lift: 0.08 divided by 0.87 equals about 9.2 percent.
5. Cost per incremental NBRx: at a hypothetical $400,000 media cost, $400,000 divided by 800 equals $500.

Compare that with the attributed figure: exposed HCPs wrote 0.95 times 10,000, or 9,500 NBRx in the window. A report that leads with 9,500 is describing activity, not impact. The gap between the two numbers is explained in attributed vs. incremental prescriptions.

A real study would add a confidence interval. If the interval on that 0.08 runs from below zero to 0.16, the honest summary is "a positive estimate we cannot yet distinguish from zero," and the next step is more time or more volume.

Common pitfalls that inflate or hide lift

• Retargeting bias. If the tactic retargets HCPs who visited the brand site, exposed prescribers were already engaged. Matched controls rarely fix this.
• Field force overlap. Reps call on high-decile prescribers, and so does media. Include call activity as a matching variable or at least as a reported covariate.
• Changing the window after seeing results. This is the most common way a weak result turns into a strong slide.
• Measuring TRx for an awareness tactic. Refills dilute the signal. Lead with NBRx.
• Too few prescribers. Small specialties produce wide intervals. Pool tactics or extend the flight rather than reporting noise.

Practical takeaway

Before your next HCP flight launches, write a one-page measurement plan that names the outcome (NBRx or another), the control type, the pre-period, and the attribution window, and get the brand lead and the measurement partner to sign off on it. Locking those four choices in advance is what makes the lift number defensible when it comes back.

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.