HCP Segmentation in Pharma: How to Build One That the Field Uses

What HCP segmentation is, and what it is for

HCP segmentation is the grouping of healthcare professionals into a small number of segments that share a commercial meaning, such as high potential but low adoption, so that each segment gets a different engagement plan. Good segmentation combines prescribing potential, current behaviour with your brand, access conditions and channel preference, and every segment maps to a decision: who the field calls, how often, through which channel and with which message.

It is often confused with targeting. Targeting answers who is worth reaching; segmentation answers how different groups should be treated. Naxion's Winter 2026 guide to segmentation and high value targeting makes the same point: the two are designed to answer different questions, are built from different inputs and are used in different ways. A brand needs both, but mixing them up produces a target list labelled with segment names that nobody acts on differently.

This guide is for brand, commercial operations and analytics leads, especially at emerging and mid-sized biopharma companies without a large analytics team. It covers the data you need (including free public data you can start with this week), a step-by-step build, a worked example with numbers, the mistakes that make field teams ignore segments, and when segmentation is not worth the effort.

The four dimensions of a useful physician segmentation

• Potential

  How much the prescriber could write in your market: patient volume in the indication, claims for the drug class, or diagnosis counts. This is the foundation and the usual basis for deciles.

  • Market volume, not your brand's volume
  • Measured on a consistent time window
  • Deciled so each decile holds a tenth of volume

• Adoption

  What the prescriber does with your brand today: share of their market volume, new patient starts, trial without repeat, or no use at all.

  • Brand share of their market
  • New to brand prescriptions
  • Recency of first and last script

• Access

  Whether their patients can get your product: payer mix, formulary position and prior authorisation burden in their area. High potential with blocked access is a different conversation.

  • Payer mix by prescriber or ZIP
  • Coverage and restrictions by plan
  • Abandonment or rejection signals

• Engagement

  How they prefer to be reached and how they have responded: rep access, email opens, webinar attendance, medical information requests.

  • Rep access and call outcomes
  • Digital engagement by channel
  • Consent and opt-out status

What data you need for HCP segmentation, including free sources

Large brands buy prescriber-level data from vendors such as IQVIA, Symphony Health and Veeva Compass, and HCP reference data from OneKey or OpenData. We compare those in Symphony Health vs IQVIA and Veeva OpenData vs IQVIA OneKey. But a smaller company can build a credible first segmentation from public data before it signs a licence, and use it to decide what to buy.

• CMS Medicare Part D Prescribers by Provider and Drug. Published free on data.cms.gov, one row per prescriber and drug, with claim counts, beneficiary counts and cost. An analysis by Morvs AI describes it as covering more than one million prescribers and roughly 25 million rows per year. CMS suppresses counts below 11, and data appears roughly a year and a half after the year it covers, so treat it as a structural view of potential, not a weekly tracker.
• NPPES NPI Registry. The public register of every National Provider Identifier, with name, practice address and taxonomy (specialty). It is the spine that joins every other source.
• Open Payments. CMS's public record of manufacturer payments and transfers of value to physicians and teaching hospitals. Useful for understanding competitor engagement and for your own compliance view; see our guide to Sunshine Act reporting.
• Your own CRM. Calls, access status, emails, events and samples from Veeva or Salesforce. This is where the engagement dimension lives.
• Specialty pharmacy and hub data, if your product runs through a limited network. It shows referrals and starts by prescriber far faster than any public source.

Two limits to plan around. Part D data covers Medicare prescription drug claims only, so prescribers who mostly treat commercially insured patients look smaller than they are. And physician-administered drugs, infusions and injections given in a clinic, are generally paid under Medicare Part B, not Part D, so for oncology, immunology infusions and similar products Part D alone will badly understate potential. In those markets you need claims data that includes medical claims.

How to build an HCP segmentation, step by step

A worked HCP decile and segment example

The numbers below are illustrative, built to show the arithmetic. They are not from a real brand.

A company launching a second-line therapy has a market of 4,000 prescribers who wrote 200,000 market claims last year. Deciling by market volume, so that each decile holds about 20,000 claims, produces a familiar shape:

• Deciles 8 to 10: 240 prescribers, 60,000 claims (6% of prescribers, 30% of volume), averaging 250 claims each.
• Deciles 5 to 7: 600 prescribers, 60,000 claims (15% of prescribers, 30% of volume), averaging 100 claims each.
• Deciles 1 to 4: 3,160 prescribers, 80,000 claims (79% of prescribers, 40% of volume), averaging about 25 claims each.

A volume-only plan would send reps to the 840 prescribers in deciles 5 to 10 and stop. Adding adoption changes the picture. Suppose that among the 240 top-decile prescribers, 90 already give the brand more than 20% of their market, 110 have tried it but sit below 5%, and 40 have never written it. Those are three different jobs:

• Champions (90): high potential, high adoption. Maintain, support with medical and patient services, avoid over-calling.
• Growth (110): high potential, low adoption. The main field focus: find out why share is low, check access, address the specific barrier.
• Not yet started (40): high potential, no use. Often an access or awareness gap; pair the rep with a market access check before increasing frequency.

Now overlay access. If 50 of the 110 Growth prescribers practise where the dominant plan restricts the product, more calls will not move share; a payer pull-through plan will. That leaves 60 Growth prescribers where frequency is the right lever. On the long tail, the 3,160 prescribers in deciles 1 to 4 hold 40% of volume, too much to ignore and too spread out for reps. That is where digital channels and inside sales earn their keep, which is the question pharma omnichannel analytics answers.

The result is a plan that points rep time at 60 named prescribers instead of 840, sends 50 to access teams, protects 90 champions from over-calling, and covers the tail digitally. That is what a segmentation is for.

Segmentation methods compared

Rules-based matrix

Every prescriber is placed by transparent rules: decile band, share band, access band. Easy to explain to reps and to audit. Its weakness is that cut-offs are judgement calls and the matrix can get crowded if you add too many dimensions.

Attitudinal research

Market research groups HCPs by beliefs and motivations, then a model projects segments onto the wider universe. Gives rich messaging insight. Expensive, and the projection to individual prescribers is often weak.

Clustering and machine learning

Statistical clustering or predictive models group prescribers by many signals at once and can refresh often. Powerful when data is rich and the team can maintain models. Risky as a first step, because the segments may be hard to name, explain or act on.

Mistakes that make the field ignore segments

When HCP segmentation is not worth doing

Segmentation is a tool for spreading limited resources across a large universe. Some situations do not need it.

• Very small treater universes. In a rare disease with a few hundred treating physicians, every one of them is known by name. Account planning per physician beats a statistical segment.
• Before you have any usable data. Pre-launch, segment on market potential and access only, and plan to rebuild once early prescribing appears. A detailed adoption-based design has nothing to work on yet.
• When the CRM cannot use it. If segments cannot be loaded onto accounts, tied to call plans and reported on, fix that first; the best segmentation in a slide deck changes nothing.
• When the real problem is sizing. If you do not yet know how many reps you need, start with pharma sales force sizing and segment afterwards.

Once segments exist, the next questions are territory alignment and call planning, which we cover in AI for call planning and territory alignment.

How Greenwolf builds HCP segmentation

We build segmentation as part of our pharma commercial analytics service, and we start with the part most projects skip: a clean HCP master keyed on NPI, linked across CRM, prescription, claims and specialty pharmacy sources, so every prescriber is the same record everywhere. On top of that we build the potential, adoption and access scores, a rules-based segment matrix agreed with the brand team, the CRM fields and call plan rules that put it to work, and a dashboard that tracks movement between segments and results against a holdout.

We work with the data licences you already hold, and for emerging companies we can build a first version from public CMS and NPPES data so you know which paid data is worth buying. Our founder started in pharma commercial operations consulting at ZS, and the team has delivered 1,000+ projects for 200+ clients over 8+ years. For the wider practice, see our specialty pharmacy analytics work.

HCP segmentation questions