Voice of the Customer in Pharma: Three Ways to Hear, Not One
Pharma needs voice of the customer more than any industry I have worked in, and collects less of it than any of them.
That is not a failure of effort. The structure of your market blocks the front door. What most companies do about it is stop listening and call the survey a listening program. You end up with a dashboard nobody acts on and a patient population you have never heard from.
There is a better answer, and it starts with admitting there are three ways to hear your customer rather than one.
Why does pharma need it most?
Because your customers meet a system you designed and cannot see.
In retail, a bad experience ends in a complaint and a refund. In your market, a bad experience ends in a patient who never starts therapy, and nobody tells you. The science worked. The clinician decided. The value evaporated somewhere between the decision and the dose, in a process your company set up and does not watch.
The scale of that silence is documented. A 2026 JAMA study summarized by Johns Hopkins found insurer rejections reached 40.7% of initial brand name attempts in 2024. Of those rejected scripts, 48.4% were never followed by a fill of that drug or anything in its class within 90 days. Almost none of those people filed a complaint with you.
The gap shows in the numbers. Deloitte's 2025 research found that only 28% of HCPs believe pharma's engagement strategies meet their needs, against 82% of life sciences executives who say they are satisfied with those same strategies. That 54 point spread is the sound of an industry not listening and not knowing it.
Meanwhile your access to the people who could tell you is shrinking. Veeva Pulse data reported by BioSpace put HCP accessibility at 45%, down from 60% eighteen months earlier. The need goes up as the door closes.
So why do you get so little of it?
Three structural blocks, and only one of them is about budget.
You often cannot speak to the patient. In most markets you may not discuss the product with her, which removes the single most informative conversation available to any other industry.
Your customer is three people. A physician chooses, a payer covers, a patient uses. Ask one and you have heard a third of the decision, usually from the party least exposed to what went wrong. I have written separately on why consumer CX does not translate to pharma, and this is the clearest consequence of it.
Your remaining access is expensive. Outside pharma, Gartner found 62% of customer service channel transitions are high effort in industries where the company can watch the transition happen. You usually cannot watch yours, and every survey you send to a physician spends a little of the 45% you have left. A listening program that burns access to produce a slide is worse than no listening program, because it costs you the thing you cannot buy back.
What are the three ways to hear your customer?
Direct, frontline and synthetic. Different fidelity, different cost, different jobs, and the discipline is knowing which one you are standing on.
Direct voice is the customer speaking for herself. Highest fidelity and most constrained. Use it sparingly, on questions nothing else can answer, and never for anything you could have observed instead.
Voice of the frontline is your field force, your hub agents, your medical information line and your patient support team reporting what they witnessed. Lower fidelity on feelings, far higher fidelity on behavior, and almost entirely unused.
Synthetic is a model standing in for a customer you cannot reach. Lowest fidelity, highest reach, and useful in a narrower band than its vendors suggest.
Most companies invest in the first, ignore the second and are being sold the third. The order of value is roughly the reverse.
Why is the frontline the most undervalued signal you own?
Because it reports behavior rather than opinion, and behavior is what you are actually trying to predict.
Your rep does not need to ask a physician how she feels about prior authorization. She watched the office manager give up on an appeal. Your hub agent does not need a survey to know which step loses people, because she has said the same sentence forty times this week. That is observed evidence, collected daily, at no incremental cost, by people you already employ.
It also arrives early. A rep meets a new payer policy weeks before any dashboard registers it. Most companies have nowhere to put that signal, so it dies in a car park between calls.
Treat it properly and it becomes an instrument. Structure what gets captured, route it somewhere a decision can be made, and close the loop back to the person who reported it, or they stop reporting. The discipline for running your field as a sensing system rather than only a delivery channel is set out in our practice note on field excellence operations.
Where does synthetic actually help?
Upstream of research, never downstream of it.
Here is the honest case. At the design stage you are not choosing between a synthetic persona and real data. You are choosing between a synthetic persona and one brand manager's assumption, formed in a meeting, unchallenged. Simulation beats assumption. That is a real gain and worth having.
Here is the honest limit. A model trained on text learns how people explain decisions, not how they make them. Your physician decides on context, habit, trust and the difficult patient she saw on Tuesday, then produces a tidy rationale when asked. The model reproduces the rationale. You get a fluent, confident account of why someone would do something they will not actually do.
That matters more in your market than in most, because predisposition rather than stated preference is what moves prescribing. It matters most of all on patients, where the pitch is strongest and there is no ground truth available to check the simulation against. The audience you most want to simulate is the one you can least validate.
So use it to generate hypotheses, pressure test a design, or reach a segment you cannot survey in time. Do not use it to produce a number you report.
How do you know which signal to trust?
Three tests, and apply them to every input before it reaches a decision.
Provenance. Can this be traced to something a human did, rather than something a human said? An abandoned appeal outranks a satisfaction score.
Predisposition or rationale. Does it capture what moved the person, or the story they told afterwards? Stated reasons are the weakest evidence in your market and the only thing a synthetic respondent can produce.
Predictive hold. Run it against a cohort whose outcome you already know. Compare what it predicted to realized conversion, time to therapy and persistence. Publish the error. Any signal that cannot be scored this way is a focus group with better grammar.
That last test is the one the industry is skipping, and it is the only one that separates a listening system from a listening theatre.
Where does this sit against measurement?
Beside it, doing a different job, and confusing the two is what broke the last generation of CX programs.
Voice of the customer tells you why. Progression measurement tells you whether. You need both, and the error was never listening too much. The error was promoting sentiment to the scorecard, where it could not carry the weight.
Keep your listening rich and your scorecard behavioral, and the whole system holds. The measures that belong on the scorecard are in how to measure customer experience in pharma, and the discipline that connects listening to realized value is Customer Excellence.
Key takeaways
- Pharma needs voice of the customer more than other industries and collects less of it, because the front door is closed by law and by structure.
- The 54 point gap between how HCPs rate your engagement and how your executives rate it is what not listening looks like from the outside.
- You have three routes, not one. Direct voice, voice of the frontline and synthetic, in descending fidelity and ascending reach.
- The frontline is the most undervalued signal you own, because it reports observed behavior rather than stated opinion and arrives weeks earlier.
- Synthetic simulation beats assumption and loses to evidence, so it belongs upstream of research and never inside a reported number.
Questions to ask your leadership team
- When did a patient last tell you something that changed a decision, and how did it reach you?
- What happens to what your field force sees, between the observation and the next planning cycle?
- Of the customer evidence in your last brand review, how much was something observed and how much was something stated?
- If you bought a synthetic audience tomorrow, what cohort would you validate it against, and who would publish the error?
- Which numbers on your scorecard are sentiment, and what would replace them?
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. Related reading: Operationalizing voice of the customer for commercial success, Bespoke, the future of experience measurement and What is value leakage in pharma?







