A third-year internist is sitting across a conference table from a health system's physician recruiter, reviewing an offer letter for a job she genuinely wants. The recruiter cites a number, describes it as competitive, and mentions, casually, that it sits "around the 60th percentile for the region."
She has no way to check that. The recruiter's organization pays for enterprise access to MGMA DataDive, described by MGMA itself as a "federally recognized source to assist in fair market value" and built from one of the industry's most detailed pools of compensation data. It shows every percentile, for her exact specialty and region, and the recruiter can see the entire distribution and choose which point in it to quote her.
She has Doximity's public compensation report, which gives specialty- and metro-level averages, useful as a rough sanity check but nowhere near granular enough to confirm or dispute a specific percentile claim. She has a group text with three former co-residents, two of whom answered vaguely and one of whom did not answer at all, because talking about pay still feels like something you are not quite supposed to do out loud, even with people you trained beside for three years.
She accepts a number she cannot independently verify, negotiated against a party who is looking at data she is structurally barred from ever seeing. The information asymmetry is not an accident of two people who happened to prepare differently. It is built into who can afford to buy the data and who cannot.
The gap this asymmetry produces, measured
The clearest evidence of what this actually costs sits in Doximity's own compensation research, and it has not been improving.
The physician gender pay gap held at 26% in 2025, according to Doximity's 2026 Physician Compensation Report, unchanged from 2024 and up from 23% in 2023. In dollar terms, men earned a model-adjusted $122,276 more per year than women, on average, across the profession. The gap is not uniform: it reached 19.7% in thoracic surgery, the largest specialty-level gap measured. And overall compensation growth itself has slowed, to just 2% between 2024 and 2025, meaning the gap is not shrinking even as the broader pool of compensation growth that might have absorbed it has stalled.
Put a number on the scale. Applying the $122,276 average gap across even 1,000 underpaid women physicians for a single year represents $122 million in foregone annual compensation, before accounting for how that gap compounds across a multi-decade career into retirement savings, Social Security credits, and total lifetime earnings. That is one year, for a fraction of the roughly one million practicing US physicians who negotiate compensation an average of three to five times across a career.
Whose data this actually is
The benchmarking infrastructure that determines physician compensation was built to serve one side of the negotiating table, and it was built well.
MGMA, Sullivan Cotter, AMGA and Gallagher collectively run the industry's compensation-benchmarking business, and their product is genuinely sophisticated: granular, current, percentile-level data broken out by specialty, geography and practice type, sold through enterprise licenses to the hospitals, health systems and large groups that can afford them. That is not a criticism of the product. It is an accurate description of who it is built for, and it is not individual physicians.
An employer negotiating with the internist in the opening scene can select whichever percentile most favors the offer being made, and has no obligation to disclose the rest of the distribution. That is not a violation of any rule. It is simply what happens when only one party in a negotiation has access to the full picture.
The physician-facing alternatives, by contrast, are built for a different purpose. Doximity's public compensation report and Medscape's annual report publish broad averages and specialty- or metro-level ranges, genuinely useful as general context, and nowhere close to the granularity an employer can see and selectively cite. Glassdoor and Salary.com carry self-reported, unverified figures that anyone can question. And the most honest channel available, informally asking former co-residents or colleagues what they actually make, remains one of the most socially awkward conversations in the profession, underused even among people who trust each other, which means even the informal workaround underperforms its potential.
The structural reason nobody has built the physician-side version
MGMA, Sullivan Cotter and AMGA are structurally employer-facing revenue businesses. Their customers are hospitals and health systems paying for enterprise licenses. A tool that gave individual physicians equivalent access would directly undercut the product they already sell, which means there is no version of their existing incentive structure that leads them to build it.
Doximity and Medscape publish aggregate compensation reports as marketing and engagement products. They drive traffic, generate press coverage, and support the platforms' broader positioning. Neither has built the granular, verified, member-queryable percentile data that would actually close the gap with what employers can see, and Doximity in particular has a business model that depends substantially on employer and recruiter relationships, relationships that benefit, at the margin, from the status quo asymmetry rather than from disrupting it.
Physician social platforms carry anecdotal compensation discussion with no verification and no systematic structure. Sermo and physician subreddits host real conversations about pay, genuinely useful in the moment, but with no way to confirm a poster's specialty, location or seniority, and no aggregation into anything a physician could cite with confidence in an actual negotiation.
The result: every party with the technical capability to build a verified, granular, physician-side compensation dataset has a direct commercial reason not to, and the party who would benefit most, the individual physician sitting across the table from a recruiter, has neither the scale nor the infrastructure to build it herself.
The structural failure: the least powerful party has the least information
State the mechanism plainly, because it is easy to describe this as simply "physicians don't talk about money," which understates what is actually happening.
The compensation-benchmarking industry sells to whoever can afford enterprise licensing: hospitals, health systems, large groups. That is, structurally, the buyer's side of every individual negotiation. The party with the least market power in any single negotiation, the individual physician, is therefore structurally the party with the least access to the exact data that negotiation depends on. This is not a symmetric information gap that happens to favor employers slightly. It is an asymmetry built directly into who the benchmarking industry's business model requires as a customer.
Independent physicians and small groups face a sharper version of the same problem: they frequently cannot afford MGMA membership at all, while the large employer they may eventually negotiate with, or compete against for referrals, holds enterprise access as a routine operating cost. The asymmetry is not just individual-versus-institution. It runs straight through independent practice as a business model.
Why now
Two developments make this the right moment to name the gap plainly and act on it.
Pay-transparency laws are spreading state by state, with Colorado, California, New York and Washington among the states now requiring salary-range disclosure in job postings. These laws are gradually normalizing the idea that compensation ranges should be visible, in a way that makes a verified, physician-facing compensation tool far more legible and far less politically fraught than it would have been five years ago.
The gender pay gap holding steady at 26% for a second consecutive year, after widening from 23% in 2023, gives the transparency case a clean, trackable annual benchmark. This is not a one-time finding that might have been a measurement artifact. It is the same instrument, run by the same organization, showing the gap persist and then plateau at a still-substantial level.
What would actually work
A verified peer-compensation graph, not another aggregate report. Real, identity-verified physicians confidentially reporting actual current compensation by specialty, metro and practice type, pooled and queryable only by other verified members, structurally mirroring what MGMA already sells to employers, but built from and owned by the population currently locked out.
Confidentiality strong enough to produce honest reporting. Compensation remains one of the most socially guarded topics in medicine even among trusted peers; a tool that cannot guarantee real confidentiality will not get honest data, and a tool built on dishonest data is worse than no tool at all.
Minimum cell sizes large enough to protect anonymity and stay legally sound. A specialty-metro combination with too few reports is both a privacy risk for the individuals in it and a potential problem under wage-information-sharing rules; any credible design needs a genuine minimum threshold before showing a percentile back to a user.
A direct connection option to consenting cohort peers, not just anonymous aggregates. The same residency-cohort mechanics that make other referral and career tools in medicine work could let a physician see verified data specifically from people she trained with, both more persuasive in an actual negotiation and more resistant to self-selection bias than a fully anonymous pool.
Explicit antitrust-safe design from the start. Physicians sharing compensation data collectively sits close to a line that federal antitrust guidance treats carefully; a credible tool has to be built from the outset around permissible aggregated, de-identified benchmarking, distinct from anything that could function as anticompetitive wage coordination.
Pricing aimed at parity with what employers already pay, not at extracting value from physicians at their most vulnerable moment. Membership dues funding baseline access, rather than charging the highest price precisely when a physician is mid-negotiation and has the least leverage, is the version of this that actually closes the gap rather than just monetizing anxiety about it.
What you can do now
If you are a physician preparing to negotiate
Ask directly what percentile the offer is benchmarked to, and ask for the source. A recruiter citing "competitive" without a specific percentile and dataset is giving you nothing you can verify; asking the direct question at least forces the number into the open.
Build a real, ongoing compensation conversation with your residency cohort now, before you need it. The informal channel works better than people expect once someone breaks the initial awkwardness, and it works far better established ahead of a negotiation than started cold during one.
Use Doximity's and Medscape's published reports as a floor, not a ceiling. They will not show you the specific percentile an employer is citing, but they will tell you whether a number is wildly outside the published range for your specialty and metro, which is worth knowing before you walk into the room.
If you lead physician recruitment or compensation for a health system
Consider disclosing the full percentile range you are working from, not just the number you are offering. In a market moving toward transparency by regulation, doing this voluntarily now is a genuine recruiting differentiator, not just a compliance gesture.
Audit your own compensation-committee practices against the same gender-gap data physicians are now reading. Doximity's 26% figure and $122,276 average gap are publicly available to every candidate you recruit; assuming they have not seen it is no longer a safe assumption.
If you negotiate physician contracts professionally
Be explicit with clients about what data you can and cannot access, and why. Physician contract-review consultants have real expertise, and also a commercial interest in the perceived scale of the asymmetry; being transparent about which specific claims are verified and which are informed estimate builds more durable trust with clients than presenting all of it with equal certainty.
Frequently asked questions
What is the physician gender pay gap in 2026? It held at 26% in 2025, unchanged from 2024 and up from 23% in 2023, according to Doximity's 2026 Physician Compensation Report, with men earning a model-adjusted $122,276 more per year than women on average and a specialty-level gap as high as 19.7% in thoracic surgery.
How can physicians see real salary data before negotiating? Currently through a combination of imperfect sources: Doximity's and Medscape's published specialty- and metro-level compensation reports, self-reported figures on Glassdoor and Salary.com, and informal conversations with colleagues, none of which approach the granular, percentile-level detail available to employers through MGMA, Sullivan Cotter or AMGA enterprise data.
Is MGMA compensation data available to individual physicians? MGMA DataDive is accessible primarily through paid organizational membership or subscription, positioned by MGMA itself as an authoritative benchmarking source for setting fair market value, and structurally built for and priced toward employers and health systems rather than individual physicians negotiating a single contract.
Why is physician compensation data hard to access for individuals? The compensation-benchmarking industry (MGMA, Sullivan Cotter, AMGA, Gallagher) is built as an employer-facing revenue business, selling enterprise-grade percentile data to hospitals and health systems; building an equally granular, verified, individual-physician-facing product would directly compete with that existing business model, which is why none of the incumbent players has built one.
How much should I negotiate a physician salary offer? There is no universal answer, but a reasonable starting approach is to compare an offer against the broadest published benchmarks available (Doximity's and Medscape's specialty- and metro-level reports) as a floor check, while recognizing that employers can see full percentile distributions individual physicians cannot access, which is why asking directly what percentile an offer represents, and from what data source, is a reasonable and increasingly normal negotiating question.
Do pay-transparency laws help physicians negotiate? They help incrementally by normalizing salary-range disclosure in job postings; Colorado, California, New York and Washington are among states with such laws in effect. They generally require posted ranges rather than full percentile-level benchmarking data, so they narrow but do not eliminate the gap between what an individual physician can see and what an employer's enterprise benchmarking access provides.
The bottom line
The internist in the opening scene did nothing wrong. She read the offer letter carefully. She checked the public compensation report. She texted people she trusted. Every step available to her, she took. None of those steps gave her what the recruiter across the table already had: the full percentile distribution for her exact specialty and region, current, granular, and purchased specifically so the employer would never have to negotiate blind.
The gap that produces is not abstract. It shows up as a 26% gender pay gap holding steady for a second year, a $122,276 average annual difference that compounds over a career into retirement savings and lifetime earnings nobody fully recovers. It shows up as independent physicians who cannot afford MGMA membership competing against employers who treat that same access as a routine operating cost. It shows up, most simply, as a number a recruiter is allowed to describe as competitive with no obligation to show what competitive actually means.
None of the organizations positioned to fix this have a business reason to. MGMA's product depends on the asymmetry. Doximity's revenue depends on the employer relationship on the other side of it. The only party with a genuine incentive to close the gap is the roughly one million physicians who negotiate blind, on average, three to five times across a career, and who have never had a way to pool what they individually know into something as useful as what the other side of the table already buys.
Somewhere in every metro area, physicians who trained together, who would tell each other honestly what they actually make if there were a safe, verified way to do it, are instead sitting across separate negotiating tables, each one guessing.
Part of a series on the missing professional infrastructure of healthcare. Previously: Helicopter Research Is a Partner-Discovery Failure
Evidence note: the core gender pay gap figures (26% in 2025, $122,276 average dollar gap, 19.7% thoracic surgery specialty gap, the 23%-to-26% 2023-2025 trend) are as published in Doximity's 2026 Physician Compensation Report, a vendor-published dataset drawn from Doximity's own physician user base rather than a government or fully independent academic source, and should be read with that caveat, though it is the most current and widely cited figure of its kind. MGMA DataDive's description as a "federally recognized source" and its characterization of its own data pool are drawn from MGMA's own product marketing. The specific claims that employers "selectively cite" favorable MGMA percentiles during negotiations, and that MGMA's dataset covers approximately 220,000 providers, are consistent with widely repeated industry commentary but could not be traced to a specific verifiable primary source in the underlying dossier research and should be treated as plausible but unverified. The $122 million illustrative calculation (1,000 physicians times the average gap) is an order-of-magnitude illustration built from the Doximity figure, not an independently published estimate. This article does not constitute negotiation, legal or financial advice; physicians evaluating a specific offer should consult current state pay-transparency requirements and, where appropriate, independent contract-review counsel.
