A medical science liaison is sitting in her car in a strip-mall parking lot outside a three-physician endocrinology practice, fifteen minutes before her next call, pulling up the KOL-mapping software her company pays for by the seat.
She is trying to answer a genuinely hard scientific question about how a new GLP-1 formulation is behaving in patients with complex comorbidities, patients who look nothing like the trial population, patients this particular practice sees all day because it is the only endocrinology group within forty miles. The software ranks candidates by publication count, conference faculty appointments, trial participation, and social-media reach. Every name on the first page practices at an academic medical center three states away and has not run a general endocrinology clinic in a decade.
The person who actually knows the answer to her question, the physician a hundred feet from her windshield who has titrated this exact drug in exactly these patients for eighteen months, does not appear on any list her software can generate. She has never published on the topic. She has never sat on a conference panel. She is, to every tool the industry has built to find expertise, invisible.
She emails the practice's front desk instead, the same channel used to schedule a sales call, and hopes someone forwards it. The industry has built an entire discovery infrastructure that can find anyone with a publication record or a following, and almost nobody who is simply, quietly, very good at the actual job.
What happens when reach substitutes for judgment
The clearest evidence of what this selection failure produces is not in a boardroom. It is in what happened when industry stopped mediating the endorsement and let visible reach select clinicians directly.
A 2024 JAMA analysis examined 28 physicians who publicly endorsed drugs or devices on a major social media platform in 2022. Every single one, all 28, had received industry general payments, totaling $1,448,083 with a mean of $51,717 per physician. 93% had received payments specifically from the manufacturer of the product they were endorsing. 61% of the promotions were sponsored testimonials, and 47% of those disclosed no compensation at all.
Now look at what the underlying expertise looked like. The median H-index was 15, a modest research footprint by academic standards. 68% had an H-index under 20. And half had no publications related to the product they were endorsing at all. These were not the field's leading experts speaking up. They were the people already inside the payment relationship who also happened to have reach, endorsing products they had no particular scholarly claim to understand better than anyone else.
That is what the market looks like when the only queryable signals are publications, trial participation, conference faculty status, and payment history, and reach is added to the mix unfiltered. It selects for exactly the wrong thing, and it does so at scale.
The payment data shows the same pattern, structurally
The endorsement scandal is a visible symptom. The underlying mechanism is bigger and less visible, and it shows up cleanly in the payment records industry itself is required to report.
Across 996,982 Medicare-billing clinicians in 109,952 practices, a 2025 Health Services Research study found that practice membership alone explained 24.8% of the variation in whether a clinician received any industry payment at all, and 60.5% of the variation in how many payments they received. Read that number again. Whether an individual clinician gets approached by industry is driven far more by which practice they happen to work in, which is a proxy for whether a sales rep already covers that territory, than by anything about the clinician's actual expertise, patient volume, or clinical judgment.
47.6% of the 996,982 clinicians studied received an industry payment in 2021. That is not a small, elite tier. It is close to half the billing physician workforce. But the distribution inside that half is the real story: over a decade, 826,313 of 1,445,944 US physicians (57.1%) received some industry payment between 2013 and 2022, totaling $12.13 billion, and the concentration at the top is extreme. The top 0.1% of recipients averaged $1,987,862. The median physician who received anything at all got $48.
Both tails are a misallocation. The top is saturated, repeat-engagement attention going to the same small pool. The bottom is a token payment, likely a meal or a minor honorarium, in exchange for essentially nothing usable as scientific input. Neither tail is where the community clinician managing real-world complexity, the one who actually knows what happens when the trial population meets an uncontrolled comorbidity, shows up at all.
Robotic surgery offers a sharper, more recent version of the same pattern. Among 268 US physicians posting publicly about robotic surgery, a 2025 JAMA Network Open study found 177, or 66.0%, had received payments from the device manufacturer whose products they were discussing. Public commentary about a device and financial dependence on that device's manufacturer are, in a majority of visible cases, the same relationship wearing two hats.
The scale of the money, and what it is not buying
Total industry payments to US physicians reached $14.67 billion in program year 2025, including $3.92 billion in general, non-research payments spread across 667,898 physicians with recorded payments (CMS Open Payments, 2026). That is a substantial, sustained investment in clinician engagement.
What it is buying, based on the evidence above, is reach concentrated among a small, repeatedly-tapped group of high-visibility clinicians, plus a very long tail of token payments to nearly half a million more, with almost no mechanism anywhere in that spend to systematically surface the clinician who runs the highest volume of a given condition in a community setting and has never published a word about it.
Why nobody built the tool that would find her
Every vendor and institution positioned to fix this has a structural reason not to.
KOL-mapping vendors (the dossier for this cluster names H1, Veeva Link, IQVIA and Komodo as the major players; H1 alone claims 200 customers across 6 continents and partnerships with 85 of the top 20 pharma companies) are paid by industry to rank against the data that is available: publications, trial roles, conference faculty status, payment history, and increasingly social reach. They are not paid to go find the clinician who is invisible to all of those signals. Building a tool that surfaced how systematically their own product misses community expertise would mean indicting their own product.
Field sales teams add relationships from rep territory coverage, which is precisely the mechanism the Health Services Research study shows is already doing most of the selecting, just informally rather than by design.
Doximity, the largest professional network available to US physicians, earns its revenue substantially from pharma reach. Its incentive is to maximize visibility and engagement for the clinicians already inside its advertising and messaging products, not to build an independent discovery layer that competes with that business.
Medical societies hold real information about who among their members treats what, in what volume, but will not broker their own membership to industry buyers; doing so would look, to the membership, exactly like what it is.
Community clinicians have no organization at all representing their availability for scientific input. Unlike academic faculty, whose institutions actively manage industry relationships as part of their job, a community internist or oncologist has no institutional apparatus advocating on their behalf for the advisory work they are qualified to do.
And the incumbent condition that makes all of this hard to disrupt: any credible fix requires independent verification of expertise that is not itself paid for by the party trying to identify experts. Nobody currently positioned in this market has an unconflicted incentive to build that.
The structural failure: geography and payment history are not expertise signals
Here is the diagnosis stated plainly. The tools industry uses to find clinical judgment are built entirely from signals that correlate with visibility, not competence: who publishes, who already gets paid, who already sits on conference faculty, who already has a following. None of those signals measure what medical affairs teams actually need, which is: who treats this condition, in what volume, in what setting, and what do that person's peers think of their judgment.
That last piece, peer attestation, is the piece no vendor has, because it cannot be purchased or scraped. It has to be volunteered by clinicians who trust the party asking for it, and no party currently asking has earned that trust, because every party currently asking is either the industry buyer itself or a vendor paid by that buyer.
The result is a selection loop that reinforces itself. Visible clinicians get asked, get paid, publish more (sometimes with industry co-authorship), and become more visible. Invisible clinicians, including the ones with the deepest real-world experience, stay invisible, because nothing in the loop has a mechanism to surface them.
What would actually work
A discovery layer built on peer attestation, not purchased signals. Verified clinicians mapped to what they actually treat, in what volume and setting, with colleagues attesting to who they would actually ask about a given question, a layer that intersects with but is not derived from publication or payment data.
Independence from the buyer. The party running the discovery layer cannot be paid by the company whose product is under discussion for the specific engagement being staffed, or the selection bias simply migrates one level up.
Explicit disclosure automated into every engagement. Every payment relationship a candidate advisor has, tied to a verified ledger rather than self-report, disclosed to the buyer and, where relevant, to the public, before the engagement happens rather than discovered after the fact.
Hard guardrails against promotional drift. No speaker-bureau placement growing out of the same pipeline, no selection linked to a clinician's own prescribing volume of the sponsor's product, no patient data anywhere in the process.
Composition specified by the buyer, filled by the network. A medical affairs team should be able to specify what a representative panel requires (practice setting mix, patient volume thresholds, geography, career stage) and have it filled from a verified pool, rather than starting from whichever names a KOL-mapping tool already surfaced.
Fair-market-value and anti-kickback compliance built into the contracting layer itself, not bolted on afterward. This is a market with real regulatory teeth, and any credible product has to treat compliance as a design constraint from day one, not a legal review step at the end.
A public account of the gap itself. An annual measurement comparing who industry actually pays for advisory input against who actually delivers the relevant care, by therapeutic area, would be the first honest public benchmark this market has ever had, and would create real pressure on every incumbent vendor.
What you can do now
If you are a community clinician
Understand that invisibility here is structural, not a reflection of your standing. The tools industry uses to find advisors were built to rank publication and payment history. If you have neither, you are, to those tools, not a low priority. You are simply not there.
If you are approached, ask what selected you. A rep territory relationship and a genuine scientific-input request are different things, and it is reasonable to ask which one you are being offered.
Keep your own record of what you actually treat, and in what volume. If a verified discovery layer for this problem does get built, that record is exactly the input it will need from you, and it is worth having ready regardless.
If you work in medical affairs
Ask your KOL-mapping vendor what signal, if any, incorporates practice volume or peer nomination rather than publications, trials, conference faculty and payment history. For most current tools, the honest answer is none, and it is worth knowing that plainly before the next advisory board is staffed from the same list as the last one.
Push back internally on staffing advisory input entirely from rep territory relationships. The Health Services Research finding that practice membership explains 60.5% of payment-count variation is a description of exactly what happens when territory coverage substitutes for a deliberate expertise search.
Ask explicitly for representativeness, and define it before the panel is built. Specify the practice-setting mix, geography and volume thresholds you actually need, rather than accepting whatever a vendor ranking returns.
If you lead compliance for a medical affairs organization
Treat undisclosed conflicts among quoted or engaged clinicians as a measurable risk, not an assumption that existing vetting works. The robotic-surgery finding, 66.0% of physicians publicly discussing a device had payments from its manufacturer, suggests the current disclosure environment is thinner than it looks.
Ask whether your fair-market-value rate-setting is calibrated to actual expertise or to a name's visibility. A rate card built around academic KOL norms may be systematically underpricing the community clinician whose input is scientifically more relevant to a given launch.
Frequently asked questions
How do pharmaceutical companies choose key opinion leaders? Primarily through KOL-mapping software that ranks clinicians on publications, clinical trial participation, conference faculty status, prior industry payments and increasingly social-media reach. A 2025 Health Services Research study found that practice membership alone explained 60.5% of the variation in how many industry payments a clinician received, indicating rep-territory coverage, not expertise, does much of the actual selecting.
How can a community physician get on a pharma advisory board? There is currently no reliable, independent pathway. Most advisory input is sourced from KOL-mapping tools that rank on publication and payment history, which structurally exclude high-volume community clinicians who have not published. Community physicians are advised to make their practice volume and clinical focus visible where they can, since no verified discovery layer for this population exists yet.
Are paid physician endorsements of drugs and devices disclosed? Not consistently. A 2024 JAMA analysis of 28 physicians endorsing products on social media in 2022 found 61% of the promotions were sponsored testimonials and 47% of those disclosed no compensation at all, despite all 28 physicians having received industry payments.
What is a digital opinion leader? An informal industry term for a clinician whose primary value to a pharmaceutical or device company is social-media reach rather than clinical or research standing. The JAMA 2024 analysis found this population had a median H-index of 15, with 68% under 20 and half having no publications related to the products they endorsed.
How much do pharma advisory boards pay physicians? Rates vary widely and are frequently benchmarked to informal fair-market-value ranges rather than public standards; the dossier evidence for this piece estimates verified community-clinician advisory programs could reasonably be priced in the $30,000 to $80,000 per program range for recruitment, verification and moderated sessions, though this is a market estimate rather than a published industry figure.
How much does industry pay physicians overall? US physicians received $14.67 billion in total industry payments in program year 2025, including $3.92 billion in general, non-research payments across 667,898 physicians (CMS Open Payments, 2026). Between 2013 and 2022, 57.1% of US physicians received some payment, with the median recipient getting $48 and the top 0.1% averaging $1,987,862.
The bottom line
The physician in the strip-mall practice is not hard to find because nobody in her field knows who she is. Every colleague within driving distance knows exactly who she is, because that is how referrals and curbside consults already work in medicine. She is hard to find because industry built its discovery machinery entirely out of signals, publications, payments, conference faculty, social reach, that measure visibility instead of judgment, and then optimized relentlessly on those signals for a decade.
The result shows up plainly in the numbers. Payment concentration at the top so extreme that 0.1% of recipients average nearly two million dollars while the median recipient gets forty-eight. A public endorsement market where every studied case involved a paid relationship and half the endorsers had no relevant publication record at all. A structural finding that which practice a clinician works at explains more of their industry payment history than anything about their actual patients.
None of this requires bad faith to explain. Medical affairs teams are not choosing to ignore the community clinician who actually knows the answer. Their tools cannot see her. That is a discovery problem, not a values problem, and discovery problems have engineering solutions: verified identity, verified practice volume, and peer attestation, assembled by a party nobody is paying to look the other way.
Until that layer exists, the industry will keep buying the loudest fifteen thousand voices in American medicine and calling it scientific exchange, while the quiet, high-volume expertise sitting in strip-mall practices across the country stays exactly where it has always been: known to every neighboring physician, and invisible to everyone with a budget.
Part of a series on the missing professional infrastructure of healthcare. Previously: The Visa-Sponsorship Penalty
Evidence note: the physician social-media endorsement findings (28 physicians, all with industry payments, median H-index 15) are from a 2024 JAMA analysis and describe a specific, likely non-representative sample of visible endorsers rather than physicians broadly. The practice-clustering finding (60.5% of payment-count variation explained by practice membership) is from a 2025 Health Services Research study covering 996,982 Medicare-billing clinicians, an observational association, not evidence of intent, and it does not establish that practice membership causes payment likelihood rather than correlating with other unmeasured factors. The robotic-surgery social-media finding (66.0% of posters had manufacturer payments) is from a 2025 JAMA Network Open study of 268 physicians, a narrow subgroup. The $12.13 billion decade total and 57.1% recipient share are from a 2024 JAMA analysis of CMS Open Payments data 2013 to 2022. Program year 2025 totals are as published directly by CMS Open Payments. H1's customer and pharma-partnership claims are vendor-published marketing figures and have not been independently verified here. Pricing estimates for a hypothetical verified advisory panel service are market estimates from the dossier, not a published industry benchmark.
