Ask three people in a pharmaceutical company how many new patients started on the brand last quarter, and you may get three different answers. All three may be calculated correctly. They will simply be answering slightly different questions, with slightly different data, and nobody will have written down which question each one answers.
This guide covers the commercial KPIs pharma teams use most often, the ways their definitions commonly differ, and how to write and govern definitions so that the same name always means the same number.
Why commercial KPIs drift
KPI drift happens when a metric with one name ends up calculated in more than one way. In pharma commercial analytics it has a few predictable causes.
Many data sources, each with its own conventions. Prescription data, specialty pharmacy and hub data, claims, payer data, CRM activity and shipment data all describe the same business from different angles. Each counts things differently, covers a different share of the market, and arrives on a different schedule.
Undocumented choices. Every calculation involves choices: how long a lookback window to use, whether to count a patient who restarts after a gap, which channels to include, how to treat free goods. These choices are usually made sensibly by an analyst under time pressure and then never written down.
Copy and adapt. A metric built for one report gets copied into another and adjusted slightly. A year later there are four versions, all called the same thing.
People move on. The analyst who built the original logic leaves, and the next person rebuilds it from the code, often getting it slightly wrong.
Vendor changes. A data supplier changes its methodology or its projection, and the historical series shifts without anyone noticing which reports were affected.
None of these are failures of competence. They are what happens when definitions live in code and in people’s heads instead of in a shared, governed place.
What a complete KPI definition includes
A useful definition has enough detail that two analysts, working independently, would produce the same number from it.
| Field | What to write | Example for NBRx |
|---|---|---|
| Name | The metric’s official name, and any aliases | New-to-brand prescriptions (NBRx) |
| Business definition | What it means, in one or two plain sentences | Patients starting the brand who have not had it in the lookback period |
| Calculation | The exact logic | Count of patients with a dispensed claim for the brand in the period and no claim for the brand in the prior 12 months |
| Unit and grain | What is counted and at what level | Patients, by week, by territory |
| Time window | The period and any lookback | Weekly; 12-month lookback |
| Data source | Where the data comes from, including the specific feed | Patient-level prescription data, retail and mail channels |
| Filters and exclusions | What is left out | Excludes free goods and samples; includes switches and add-ons |
| Owner | The person accountable for the definition | Director, commercial analytics |
| Refresh | How often it updates | Weekly, Tuesday |
| Version and effective date | The current version and when it took effect | Version 3, effective January 1 |
| Caveats | Known limitations | Mail-order coverage is partial in some regions |
Our free data dictionary template includes a metrics sheet with these fields.
Prescription metrics
These are the core volume metrics for most retail and many specialty brands.
| KPI | Common definition | Where definitions usually differ |
|---|---|---|
| TRx (total prescriptions) | All prescriptions dispensed in the period, new and refills | Whether counts are normalized to a standard days of supply; which channels are included; projected versus observed counts |
| NRx (new prescriptions) | Prescriptions dispensed under a new prescription number | Often confused with NBRx. NRx includes new prescriptions for patients already on the drug, such as a renewal after refills run out |
| NBRx (new-to-brand) | Patients starting the brand who have not had it in a lookback period | Lookback length (6, 12 or 24 months); whether patients restarting after a gap count as new; whether switches and add-ons are split out |
| Market share | The brand’s volume as a share of the market basket | Which products are in the basket; whether share is measured in TRx, NBRx, patients or days of therapy; volume versus value |
Two points cause most of the confusion.
NRx is not new patients. NRx counts new prescription numbers, and a patient who has been on a drug for years generates a new prescription number whenever their old prescription runs out. NBRx is the measure of new patients. Use the right one for the question being asked.
Share depends entirely on the basket. A brand’s share in “all treatments for the condition” can be very different from its share in “branded biologics in second line.” Both are legitimate. They need different names, or at least a written definition that makes the basket explicit.
Patient metrics
Specialty brands and brands with complex patient journeys rely heavily on patient-level metrics, which have even more room for variation.
| KPI | Common definition | Where definitions usually differ |
|---|---|---|
| Patient starts | Patients who receive their first dispense of the brand | Whether a start is counted at referral, at benefits verification, at approval or at first dispense |
| Persistence | Share of patients still on therapy a set time after starting | The allowed gap before a patient is counted as discontinued (often 30, 60 or 90 days); whether a switch counts as discontinuation |
| Adherence (PDC) | Share of days in a period on which the patient had medication available | The measurement period; how overlapping fills and hospital stays are handled |
| Adherence (MPR) | Total days of supply dispensed divided by days in the period | Can exceed 100 percent, and is often capped; less conservative than PDC |
| Abandonment rate | Share of approved prescriptions never collected by the patient | Whether rejected claims are included; the time allowed before counting as abandoned |
| Time to therapy | Days from prescription or referral to first dispense | The starting point; how cancelled referrals are treated |
For adherence, many organizations follow the Pharmacy Quality Alliance, whose measures treat a proportion of days covered of 80 percent or more as adherent for most drug classes, with a higher threshold of 90 percent for antiretrovirals. Whatever threshold your team uses, write it down.
Field and promotional metrics
| KPI | Common definition | Where definitions usually differ |
|---|---|---|
| Reach | Share of target customers with at least one interaction in the period | Which target list; which interaction types count (in-person, remote, email, events) |
| Frequency | Average interactions per reached customer | Same as reach; whether repeated same-day contacts count once |
| Call activity | Number of customer interactions | What counts as an interaction; whether medical interactions are excluded (they usually should be) |
| Share of voice | The brand’s share of promotional activity in its market | Measured by interactions, by spend or by impressions; which competitors are included |
Access and pricing metrics
| KPI | Common definition | Where definitions usually differ |
|---|---|---|
| Covered lives | People in health plans that cover the brand | Which access levels count as covered (for example, whether coverage with prior authorization counts); which plan types are included |
| Formulary access | Share of lives with access at a stated level | The level used: preferred, covered with restrictions, or any coverage |
| Pull-through | How well favorable access converts into prescriptions | Measured as share in covered versus uncovered plans, or as change after a formulary win |
| WAC | Wholesale acquisition cost, the list price to wholesalers | Usually unambiguous, but check the effective date of price changes |
| Gross-to-net | The difference between gross sales at list price and net sales after rebates, discounts and other deductions | Which deductions are included; accrual versus actual; product versus portfolio level |
Gross-to-net figures are often the most sensitive metrics in the company. They also tend to be the least consistently defined, because finance, access and brand teams each look at them for different purposes.
A worked example: two NBRx numbers
A brand team reports that NBRx grew eight percent last quarter. The commercial analytics dashboard shows three percent. Leadership asks which is right.
An analyst investigates and finds:
- The brand team’s figure uses a six-month lookback. Any patient who had not filled the brand in six months counts as new.
- The analytics figure uses a twelve-month lookback. A patient has to have been off the brand for a full year to count as new.
- The brand had a supply interruption eight months ago. Many patients paused therapy and restarted last quarter.
Under the six-month rule, those restarting patients count as new to brand. Under the twelve-month rule, they do not. Both numbers are calculated correctly. They answer different questions: “how many patients started or restarted?” and “how many genuinely new patients did we win?”
The fix is not to decide that one team was wrong. It is to agree which definition belongs on the leadership scorecard, document it, name the other one differently if it is still needed (for example, “starts including restarts”), and record the decision so the question does not come back next quarter.
Without written definitions, this investigation typically takes days. With them, it takes minutes, because the difference is visible in the definitions themselves.
Keeping definitions from drifting
Writing definitions once is not enough. They need light, continuous governance.
One home for definitions. Every official KPI definition lives in one shared place that everyone can read: a business glossary or data dictionary, not a slide in someone’s drive.
A single owner per KPI. Each metric has one named owner who decides what it means. Others can propose changes. The owner decides.
Simple change control. Changes to a definition go through a short review: what is changing, why, which reports are affected, and from when. The review can be a fifteen-minute conversation. What matters is that it happens and is recorded.
Versions and effective dates. When a definition changes, the old version is kept, and the new one has an effective date. Historical comparisons then remain meaningful.
Lineage from report to definition. Every report and dashboard should state which definition it uses, and ideally which version. Data lineage makes it possible to trace a figure on a dashboard back to the tables and rules that produced it.
Regular review. Once or twice a year, review the full list. Retire metrics nobody uses, merge near-duplicates, and confirm that owners are still in their roles.
Where to start
Do not try to define every metric at once. Start with the KPIs on the leadership scorecard, because those are the ones whose disagreements are most visible and most expensive. Document each one using the fields above, agree the owner, and link the scorecard to the definitions. Then expand to the metrics used in brand reviews, and then to the rest.
Most teams find that a dozen well-documented definitions settle the majority of their recurring arguments.
Further reading
- The Pharmacy Quality Alliance adherence measures describe the proportion of days covered method and thresholds used across much of US healthcare.
- Our commercial data governance page describes how Clarity keeps definitions, lineage and ownership in one place.
- The glossary defines the data governance and commercial terms used here.