Retail Purchase Signal at 1/9th the Cost of Rx Claims

The question worth answering

DTC pharma buys carry a quiet assumption: the more clinical the audience data, the better the outcome. Rx claims audiences, health-verified segments, and modeled patient cohorts command CPM premiums of two to three times what commerce data costs, on the logic that a person identified through the healthcare system is a more valuable impression than a person identified through a shopping cart.

The problem is that this assumption almost never gets a fair test. Partners run in different months, on different flight lengths, measured by different vendors on different definitions of a reached patient. Any comparison you assemble afterward is contaminated before you start.

This campaign accidentally produced the clean version. A pediatric immunization brand ran ten audience data partners in the same December, in the same buy, against the same brief, with every line measured by one third-party Rx outcomes platform on one verified-patient definition. Same month, same platform, same yardstick. That controls for flight-length distortion and vendor methodology, so the ranking reflects the data, not the media plan around it.

Inmar entered that field late, with four shopper segments built on retail purchase signal, categories like baby formula, nursing and feeding, and diapers, running on just 2.5% of December media.

What the data showed

Inmar finished first of ten partners on cost per verified patient reached, and its three feeding segments took the top three individual line placements out of 43 in market. At 347 verified patients per $1,000 of media, it returned 4.1 times the patient volume per dollar of the campaign it sat inside. And because the budget delivered at 1.24x frequency against a 2.75x campaign average, that money bought new households rather than repeat impressions.

The full December ranking, every partner on the same measurement:

The shape of that table is the story. The two cheapest signals, retail purchase and contextual intent, cluster at the top. The most clinical signals, claims-based and Rx claims audiences, cluster at the bottom, at five to nine times Inmar's cost per patient. The premium data did not fail to find patients. It failed to find them at a price the reach math could survive.

Retail against retail, on identical segments

The sharpest comparison in the buy removes even the data-category variable. A second retail purchase partner ran the same four segment concepts in the same month, so the only differences left are the underlying purchase data and the media it cleared on. Inmar paid 35% more per thousand impressions for its media and still cleared the matched partner on three of four segments:

Across the three feeding segments, Inmar averaged $2.78 per patient against the matched partner's $5.09, a 45% advantage, while also delivering more patients per 1,000 people reached (2.04 vs 1.74). Diapers lost cleanly, 69% more expensive per patient, and it stays lost: it is excluded from the go-forward segment recommendation. A test that cannot produce a loser is not a test.

Why it worked

Purchase signal is broad, and broad is what reach efficiency wants. Feeding-category shoppers are a large, addressable pool. Inmar's segments delivered at 1.19x frequency where the matched retail partner needed 2.3x to 2.8x to spend the same brief. When almost every impression lands on a new household, cost per person reached collapses.

Low data drag on the CPM. Inmar's lines cleared at a $4.67 CPM against a $9.52 campaign average, while the claims-based audiences ran $11.01 to $11.45. That data premium is baked into every impression before it delivers anything, and in this buy it is the gap the clinical audiences never recovered.

Normal patient density, bought cheaply. Inmar delivered 2.02 verified patients per 1,000 consumers reached against a 2.23 campaign average. The advantage is cost, not concentration: the retail segments found an ordinary rate of real patients for a fraction of what the campaign paid elsewhere to find them.

What I’d take to the next plan

Treat audience data as a portfolio with jobs, not a quality ladder. This read says retail purchase signal is a reach engine: it finds real patients at scale for single-digit dollars. It does not say claims data is useless. The higher quality indices on the clinical audiences are exactly what you want when the job is precision, suppression, or measurement seeding rather than efficient reach. The mistake is paying precision prices for a reach job.

Demand same-month, same-platform reads before reallocating. The only reason this ranking is trustworthy is that every partner was measured on one December window, one platform, one verified-patient definition. Partner scorecards assembled across different flights and vendors will quietly reward whoever had the longest measurement window. If the comparison is not structurally clean, fix the structure before moving budget.

Small tests can produce decision-grade answers. Inmar's entire flight was $8,151, 2.5% of one month. Because the measurement design was clean, that was enough to rank first of ten, win three of four matched segments, and generate a specific recommendation, including a segment to cut. The cost of learning is low when the read is designed before the dollars move.

Report the caveats with the win. The frequency mechanics, the quality-index gap, and the reporting threshold all made it into the client-facing readout, because a result that survives its own footnotes is the only kind that earns a bigger test. The broader measurement architecture behind readouts like this is covered in the Pharma Programmatic Measurement Framework, and the planning benchmarks it feeds live in Pharma Programmatic Benchmarks.