Customer Intelligence in Pharma: Context, Orchestration, Progression

Customer intelligence in pharma is the capability that puts customer reality into a commercial decision before the decision is made. Most organizations hold pieces of it and very few hold the whole thing. The pieces are usually a survey program, a field reporting process and, increasingly, an AI modelled audience, each run by a different function against a different question.

Institutions tend to listen in the manner that is easiest to administer rather than in the manner that is hardest to ignore. A survey can be commissioned, scheduled and reported, so it becomes the instrument of record. What customer-facing colleagues observe every week arrives without a template, so it rarely reaches a decision intact.

In pharma that pattern carries a specific cost, because the decisions in question govern whether a patient ever starts therapy. The clinical decision is the pivot point of the therapeutic value chain, and it depends on the commercial model around it. Intelligence is therefore part of that value chain rather than a service alongside it.

Context, Orchestration and Progression name AI's role in a commercial system

I describe AI's role in a commercial system as three things, which are Context, Orchestration and Progression. Context is what the system knows about a customer's actual situation. Orchestration is what the system does with that knowledge across channels and functions. Progression is whether the customer moved, which is the only evidence that the first two worked.

Taken in that order the three become a method rather than a technology agenda. Most commercial AI investment so far has gone into the middle one, because orchestration has products attached to it. Context is harder to buy and Progression is harder to admit, which is roughly why programs can improve visibly and change very little.

Context comes from three different listening instruments

Context is produced by listening, and the clearest evidence that pharma's listening is not working sits in the gap between two scores. Deloitte 2025 research found only 28 percent of HCPs believe pharma's engagement strategies meet their needs, against 82 percent of life sciences executives who say they are satisfied with those same strategies. A gap of that size is rarely a measurement error.

Two parties are scoring two different things, and one of them is scoring its own activity. That happens when the instruments in use report what the organization did rather than what the customer met. Three instruments are available, they are genuinely complementary, and each one is precise about something the other two are vague about.

Direct voice of the customer

Direct voice of the customer is what a prescriber, a patient or a caregiver reports when asked. Done well it can be reliable about sentiment, preference and stated intent, and it carries the authority of the customer's own language. A sound approach to voice of the customer in pharma treats those words as evidence of how something felt rather than as evidence of what happened.

Its limits are structural rather than methodological, which means better questionnaires do not remove them. Asking happens on a schedule, so the answer describes a past state by the time anyone reads it. Recall compresses, so the moment a case stalled gets remembered as a general impression. A survey may establish that an authorization process is frustrating while telling you nothing about which step broke on which account.

Voice of the frontline

Voice of the frontline is what customer-facing colleagues observe in the ordinary course of doing their work. My argument for treating it as the superior signal rests on four properties. It is observed rather than recalled, account specific rather than sampled, current rather than lagged, and already attached to the moment progression broke. Those four together make it the most unimpeachable and actionable intelligence in the commercial system.

The frontline is one of pharma's most underutilized intelligence systems, encountering barriers long before dashboards detect them. A field colleague often knows which practice has stopped starting patients, and why, several weeks before a prescribing report registers the change. The organization usually holds that knowledge somewhere and offers it no route to travel.

None of that reflects anyone's competence. Field teams report what they have been given a form to report, and those forms were designed to capture activity rather than barriers. Treating voice of the frontline as intelligence rather than as administration is a design decision, which means it can be made.

Synthetic personas

A synthetic persona is a modelled customer, assembled from available evidence and used to rehearse a question at a speed and scale fieldwork cannot reach. Run well it compresses weeks of exploratory work into an afternoon and surfaces hypotheses nobody in the room had thought to test. That is a genuine contribution and a narrow one.

The limit deserves saying plainly rather than in a footnote. A synthetic persona is a hypothesis generator, never a source of truth, and its value collapses the moment it is validated against itself rather than against observed reality. The failure mode is subtle enough to have earned a name, and I have set out the calibration trap argument in full elsewhere rather than restating it here.

The honest framing of synthetic personas in pharma is that they are an instrument inside a method rather than a capability in themselves. The instrument is already commoditising, which is ordinary for tooling of any kind. The method that decides what to ask it and what to do with the answer is the part worth owning.

The three instruments compound only when they run in order

None of the three is sufficient alone, and almost every organization runs one of them. Synthetic personas generate the hypothesis. Direct voice tests whether people recognise it. The frontline confirms whether it is happening.

Run in that sequence the three compound, because each answers a question the one before it could not. A hypothesis no customer recognises gets discarded cheaply. A hypothesis customers recognise and the frontline cannot find is a perception problem rather than an operating one. A hypothesis all three confirm is a barrier with a location, an owner and a cost attached.

Run in isolation each produces a defensible artifact and no decision. The persona study becomes a deck, the survey becomes a tracker, and the field observation becomes a call note. Every one of those can be well made by capable people. None of them on its own tells a commercial leader what to change next week.

Orchestration is where context becomes continuous progress

Context that never reaches an action is a research output with a longer name. Orchestration is the half of the method where what the system knows changes what the system does. Channel coordination makes outbound activity consistent, while journey coordination makes a customer's progress continuous. Context is what the second one requires and the first one does without, which is the distinction most omnichannel programs are never forced to confront.

A company can coordinate every channel on almost no context at all, because consistency is a property of the message rather than of the customer's situation. Continuity is a different achievement, since finishing a case that started elsewhere requires knowing what has already happened to whom. Recognition and memory are context problems that get routinely misdiagnosed as orchestration problems.

This is the point where the three instruments earn their keep operationally. A barrier that direct voice registered as frustration and the frontline located on a named account can be routed to someone with the authority to clear it. Without that context, orchestration can only sequence what it was already planning to send.

Progression is where the method is proved or discarded

Progression measures whether a customer moved, using a stage ladder rather than an activity count. The stages run Scripts Written, Filled, Therapy Started, 90-Day and Persistence, organised under the Path to Prescribe, the Path to Fulfill and the Path to Adhere. An intelligence capability is judged by whether a named stage improved after it reported something.

The health measure of the whole method is the Transfer Coefficient. It is the proportion of meaningful frontline intelligence that survives the journey from customer observation to consequential organizational action and measured effect. Few organizations have calculated it, and those that estimate it tend to find the answer uncomfortable.

A method with a low Transfer Coefficient is a research function, not an intelligence capability. The distinction is practical rather than semantic, since the two cost similar money and return quite different things. Written prescriptions are intent rather than realized value, and the same logic applies upstream to insight. An observation that never reached a decision was intent as well.

Customer consciousness is what the method is for

Customer consciousness is the degree to which customer reality is present in decisions before they are made. It describes a property of decisions rather than a sentiment about customers, which separates it from the warmer vocabulary it often gets grouped with. Raising that degree is the purpose of Customer Excellence, and a working intelligence method is how the raising actually happens.

Customer Consciousness sits as the hub of the flywheel, the state the three service lines produce together. Brand work, field work and enterprise culture work each raise it from a different direction. A listening system is the mechanism underneath all three, because none of them can read a customer's world without one.

Can a quarterly cadence serve a continuous ecosystem?

Pharma's commercial intelligence has historically run on a quarterly or annual cadence, against a healthcare ecosystem that now moves continuously. Access policies change, formularies shift, practice patterns reorganise, and patient expectations get reset by services from outside this industry entirely. A market research cycle that reports in twelve weeks is describing a world that has already changed.

The obvious response is to point AI at the cadence and make the cycle faster. That response deserves a plain answer rather than a hedged one. AI will not make a fragmented commercial model customer-centric. It makes whatever operating system exists faster and more scalable.

The consequence of that is uncomfortable and worth stating directly. AI raises the stakes on the operating model rather than substituting for it. An organization that accelerates a quarterly cadence without repairing what it listens to will simply be wrong faster, with more confidence and better charts behind it.

FieldOS is where the loop gets operationalised

A method needs somewhere to live, and in this practice that place is FieldOS. FieldOS is an operating blueprint for field intelligence and activation, defining the direction an organization aspires to take its field operating model. Components are tailored per organization rather than installed as a fixed product.

The loop runs six stages, which are Capture Signal, Classify Barrier, Assign Owner, Intervene, Measure Progression, then Learn and Improve. AI sits at the centre of that loop supporting every stage rather than occupying one of its own. Reading the field impact operating system alongside this page gives the mechanics, where this page gives the method.

The ambition behind all of it is better listening rather than better pre-call planning. Most field AI on offer improves the talking, through next best action, message refinement, call preparation and territory optimisation. Improving the listening is the harder build, and it is largely the one that moves a progression number.

Key Takeaways

• Customer intelligence in pharma is the capability that puts customer reality into a commercial decision before that decision is made.
• AI's role in the commercial system is Context, Orchestration and Progression, and context is the part almost nobody has built properly.
• Three listening instruments produce context, and direct voice, frontline observation and synthetic personas are each precise about a different thing.
• Synthetic personas generate the hypothesis, direct voice tests recognition, and the frontline confirms whether it is happening, which only compounds in that order.
• The Transfer Coefficient measures how much frontline intelligence survives to consequential action, and a low one marks a research function rather than an intelligence capability.
• AI makes whatever operating system exists faster and more scalable, so accelerating a broken listening cadence produces faster error.

Diagnostic Questions to Consider

1. Name which of the three listening instruments your organization actually runs, and which decisions each one has changed in the past year.
2. State the elapsed time between a field colleague observing a barrier and someone with authority acting on it.
3. Estimate your Transfer Coefficient by tracing ten recent frontline observations through to a measured effect.
4. Identify whether your last synthetic audience exercise was validated against observed reality or against its own prior output.
5. Report which stage of the progression ladder improved as a result of your most recent customer research investment.

Closing Reflection

The instruments in this method are not equally fashionable, and the least fashionable one is the most valuable. Synthetic audiences attract attention because they are new, visible and demonstrable in a meeting. Frontline listening attracts far less, because it asks for process change rather than procurement, and it produces findings that implicate the operating model.

That ordering of attention is worth resisting. The work I have seen change a progression number has almost never started with a better model of the customer. It has started with a route by which what the field already knew could reach someone able to act on it, quickly enough for the action to still matter.

Exceptional science deserves an exceptional commercial system, and a commercial system is only as conscious as what it listens to. The healthcare ecosystem is now continuous, which makes a quarterly understanding of it a liability rather than a discipline. The organizations that take this seriously will own a method rather than a tool, since tools change hands and methods compound.

About the Author

Wayne Simmons is the founder of The Customer Excellence AGENCY and the author of The Customer Excellence Enterprise (Wiley, 2024). He is founding faculty of the MS in Customer Experience Management at Michigan State University's Broad College of Business. He led global customer excellence in Pfizer's first Chief Marketing Organization and in Bayer's Customer Powerhouse.

The Customer Excellence Agency: Advancing the Pursuit of Excellence in Service of Science.