The work, and what it could not prove
Published research where it exists. Anonymized studies everywhere else. Caveats included, because a study without them is marketing. Each study names the data it ran on.
We wrote the research that defined the problem, then built the fix
Two industry papers, three years apart, both published by the Point of Care Marketing Association and free to read. The first named what was broken. The second proved a way to repair it, tested by someone else.
- 2023The problem
Is Your Point of Care Marketing Campaign Measurement On Point?
Author: Joy Joseph, Syneractiv. Research sponsor and chief editor: Nicole Divinagracia, POCMA. Five months of research: an industry-wide survey, expert interviews, and empirical benchmark analysis, with contributors from Merck, Novartis, ZS, Publicis Health Media, Phreesia, PatientPoint, CheckedUp, ConnectiveRx, Definitive Healthcare and others.
The paper produced seven findings. The one that mattered most became the next three years of work:
Finding #5 — POC input data in Marketing Mix Models may not capture the full depth and breadth of POC activity. MMM studies, which use only counts of HCPs with POC in any given week, will treat an office seeing 100 patients vs. 1,000 patients equally.
It also quantified the gap that had made POC budgets impossible to defend: test-vs-control benchmarks ran about 1.5x the lift that MMM implied for the same channel. Two methods, two answers, no way to tell which was wrong.
Read the 2023 paper ↗ - 2025The fix, tested
Unlocking Point of Care Marketing's True Impact
An industry research collaboration between POCMA, Trinity Life Sciences, Syneractiv and MedFuse.
Recommendation #4 of the 2023 paper said to replace coarse POC inputs with granular patient traffic. The Patient Reach Index is that recommendation built: breadth, meaning how many offices carried the placement, times depth, meaning qualified patient traffic through each one, computed at HCP-week granularity. The index is source-agnostic — it needs patient traffic per office, not a particular vendor.
Trinity Life Sciences tested it independently across 8,000 HCPs over 26 weeks, with 3,500 receiving POC campaigns.
4.4×Higher POC lift versus the binary model+51%Improvement in explanatory power−22%Reduction in prediction errorThe result that mattered was none of those. The PRI-based model landed close to an independently run matched-pair test-versus-control study. Two methods with entirely different failure modes agreed — closing the 1.5x gap the 2023 paper had identified. The binary model agreed with neither.
The caveat, stated in the paper itselfRead the 2025 paper ↗Single-brand study. It establishes that the input works in one case, not that the magnitude generalizes across therapeutic areas or brand life stages. A consortium study across multiple brands is underway to test exactly that.
Both papers are published by the Point of Care Marketing Association and publicly available. Neither is gated, and neither requires talking to us first.
Point-of-care placements drove first-time starts in a crowded specialty market
A specialty brand was buying point-of-care placements in a diagnostic-imaging network. The category was crowded with established competitors. The brand team could not establish whether the placements were producing anything the brand would not have gotten anyway, which is the only question that matters and the one the platform's own reporting structurally cannot answer.
Patients with an imaging visit at a campaign-network location formed the exposed cohort. Patients visiting non-network radiology providers in the same window formed the control candidate pool. Cohorts were propensity-matched on specialty, state, pre-period prescribing activity, and prescribing trajectory.
Pre and post windows were held symmetric at eight months each. That detail is not decoration: when the post window runs longer than the pre, conversion and script-count metrics inflate for reasons that have nothing to do with the campaign. Symmetric windows remove the bias.
Analysis was patient-level difference-in-differences with ANCOVA adjustment against MedFuse prescription and medical claims.
Exposed patients were roughly 17% more likely to start the brand for the first time than matched controls. New-to-brand was the signal, not total volume, which is what you would expect from a channel reaching patients at the moment of a treatment decision rather than reinforcing existing prescribing. Lift held in the same direction across every prescriber specialty in the analysis, including specialties outside the brand's core target.
The brand team reprioritized toward prescribers with a positive pre-campaign prescribing slope, who converted at 68.3% against 64.3% for flat or declining ones. Trajectory became a targeting input alongside specialty and volume.
Prescriber linkage covered roughly 57% of network locations. Absolute patient counts were projected to full-network scale using a coverage factor derived from two independent methods. Lift ratios were not projected, because they are scale-independent and projecting them would have manufactured precision that did not exist.
The exposed cohort was geographically concentrated, reflecting the network's footprint rather than the brand's national market. The read describes those markets and is not a national estimate.
This was a single campaign wave. It establishes that lift occurred. It does not establish a response curve, and it cannot tell you what the tenth impression is worth.
Proving a venue is on-pathway before the money is spent
Media plans routinely place point-of-care advertising in venues on the theory that the right patients pass through them. The theory is rarely tested. A brand team wanted to know, before committing budget, whether a diagnostic-imaging network actually sat on the pathway to the oncologists they were targeting.
MedFuse claims were used to trace patient flow in both directions. Downstream: of patients visiting the network, what share subsequently reached a targeted oncologist? Upstream: of patients seen by targeted oncologists, what share also visited the network?
Bidirectionality is the design choice that makes this worth doing. Downstream flow implies the venue reaches patients before the treatment decision, which is an awareness argument. Upstream flow implies it reaches patients already in active care, which is a different audience and a different creative brief. A one-directional analysis would have supported whichever story the plan already wanted.
Both directions confirmed the venue was on-pathway. Roughly one in four targeted oncologists had patients who visited the network. Around 78% of active locations showed at least one patient reaching a target prescriber, so connectivity was broad rather than concentrated in a handful of sites. Timing was fast: the majority of pathway patients reached a target oncologist within 30 days, and over 80% within 90.
The timing distribution changed the plan. A clear majority of venue visits occurred after the oncologist visit, not before, which meant the placement was largely reaching patients in follow-up imaging ordered as part of an oncology workup. That is a workflow venue reaching patients already in treatment, not a top-of-funnel venue reaching them pre-consultation. Different audience, different message. The plan had assumed the opposite.
The analysis establishes that the venue is on-pathway. It does not establish that advertising there causes anything. Pathway validation is a necessary condition for a placement to work, not evidence that it worked. Measuring that requires the lift study, which is a separate design and appears above.
Exposure at the venue was ambient and not filtered by indication, so the pathway rate describes all patients passing through, not a qualified audience.
Location-level claims linkage was incomplete, with roughly half of network NPIs showing claims activity in the period, most likely because centers bill through a single NPI despite multiple rendering providers. Coverage was treated as a floor on connectivity, not a ceiling.
Why every study on this page says what it could not prove
A measurement partner who only reports findings that flatter the campaign is not measuring. They are reporting. The value of an independent read comes entirely from the possibility that it comes back null, or smaller than hoped, or true only in the markets where the data was thick enough to support the claim.
We would rather show you the boundaries of what we proved
than let you find
them yourself.