A promotional effectiveness tracker's headline sample size is usually the first number a brand team looks at, and often the last thing anyone questions. It can be a strong number and still not mean what it appears to mean, either because it quietly includes interactions that aren't comparable to a live rep visit, or because it's been smoothed across months in a way that hides what's happening right now. Neither problem shows up unless you ask about it directly.
Below are 5 questions worth asking about any promotional effectiveness tracker, including your current one. Most trackers get the operational basics right: recruiting physicians every week, fielding surveys on schedule, and turning responses into a dashboard on time. These questions aren't about whether that machinery runs. They're about what happens to the numbers once they're inside it, and they matter most to commercial insights, analytics, and commercial effectiveness teams who report those numbers upward.
Not every HCP touchpoint is the same kind of event. A live rep detail, whether in person or by video, is a two-way conversation the physician actively participated in. An informative email or a pre-recorded video is a one-way broadcast the physician may not have opened at all. Grouping both into one sample size, sometimes called personal promotion (PP) and non-personal promotion (NPP), makes the number look bigger without making it more meaningful.
Here's how that plays out in practice. Consider a tracker that reports a monthly sample of 63 "details" for a brand: 43 in-person visits and 6 phone or video calls (49 total, genuine personal promotion), plus 12 informative emails and 2 pre-recorded videos (14 total, non-personal promotion). That's roughly 22% of the reported sample coming from touchpoints with no rep involved at all. The real personal-promotion sample is 49, not 63.
The bigger risk is what that inflated number gets used for. If a tracker applies the full sample, personal and non-personal combined, to metrics like Share of Voice, Message Recall, or Intent to Prescribe, every one of those numbers is measuring something other than what it claims to measure. Message recall from an email open is not the same signal as message recall from a live conversation, and averaging them together doesn't produce a richer number. It produces a wrong one.
This is a methodology question, not a data quality one, and it's worth asking plainly. A standalone monthly or weekly readout calculates each data point from that period's responses only. A rolling 12-week (R12W) readout calculates each data point from the prior 12 weeks pooled together, then slides that window forward one week at a time. Both are legitimate techniques. They answer different questions, and a report that doesn't say which one it's using is answering a question you didn't ask.
The arithmetic is worth sitting with. A tracker recruiting roughly 10 HCPs a week and reporting on an R12W basis will show a headline sample of around 120, twelve weeks' worth pooled together. A tracker recruiting 120 HCPs in a single month and reporting that as a standalone figure shows the same headline number, 120. They look identical on a slide. One of them reflects that month's actual field activity. The other reflects a three-month blend where the current week contributes roughly 8% of the total figure. Ask which one you're looking at.
This is the practical test of question 2. A rolling window doesn't just report differently, it responds differently to change. If a competitor launches, a label updates, or a formulary shifts, a standalone monthly readout reflects that shift in the very next reporting period. A 12-week rolling readout absorbs the same event across 11 prior weeks of unrelated history, which means the shift can take 2 to 3 months to become visible, and it can look like "minimal movement" the whole time it's happening.
That lag isn't limited to Share of Voice. Once a rolling window is applied, every downstream metric inherits the same delay: message recall, intent to prescribe, rep value, and brand perception all smooth out real change at the exact moment a brand team most needs to see it. Ask your tracker directly how a market event would show up, and how long it would take.
A sample size is only as useful as the physicians inside it. A tracker pulling from a syndicated panel shared across multiple brands can hit an impressive-looking headline number while actually reaching very few of the physicians on your specific target list. "On-target" needs a definition you can check, not just a percentage you're told to trust. Ask whether the panel is built from your call plan specifically, and whether the reported on-target rate would hold up if you audited it against your own target list.
This is also where market basket accuracy belongs in the conversation. A tracker that includes products not actually indicated for the disease stage being measured, or that can't separate early-stage from late-stage messaging when a single rep visit covers both, is diluting the same sample size a second way, on top of whatever personal-versus-non-personal blending is already happening.
A clean sample and an honest readout method solve the input problem. They don't solve the output problem. A dashboard of raw KPIs, Share of Voice, message recall, rep value, tells a brand team what happened. It doesn't tell them what to do next unless someone is running segmentations, driver models, or subgroup regressions on top of it. Ask what analysis comes with the numbers, and ask to see an example before you commit to a renewal.
This question sits well past sample size and readout method, but it's the one that determines whether the first four even matter. A perfectly clean, correctly-labeled sample that only ever gets reported as a flat KPI table is still leaving most of its value on the table, a version of the same "so what" gap explored in ZoomRx's look at whether a pharma brand is still promotion sensitive.
None of these five questions are about replacing your tracker. They're about knowing what the number in front of you is actually built from before you act on it. The same standard applies upstream of the tracker too. A visit that clears the bar physicians actually set, the subject of ZoomRx's research on what earns a physician's time in 2026, and a visit that leaves nothing behind afterward, can both register as identical rows in a tracker that only counts occurrences. The five questions above are how you check whether yours does more than count.
This data is drawn from ZoomRx's methodology briefs, "The Sample Size Illusion" and "The Rolling Sample Problem." Both are available as ungated resources below: