Flatiron Health

Flatiron has two completely separate buyers who share nothing except the word "oncology."

Buyer A: A VP of Real-World Evidence or Head of Regulatory Strategy at a top-20 pharma company who needs oncology-specific patient-level data (not claims data, not chart abstraction) for regulatory submissions, HTA filings, and clinical trial design. This buyer's Monday-morning problem is "the FDA wants real-world evidence for our supplemental indication and IQVIA's claims data doesn't have treatment response."

Buyer B: A practice administrator or medical director at a community oncology center who needs an EHR built for regimen-based workflows, clinical trial matching, and prior auth. This buyer's Monday-morning problem is "Epic Beacon doesn't think like an oncologist."

These look like two completely different buyers. But the differentiated capability they both buy is the same: oncology data captured at the point of care, not reconstructed from claims or charts after the fact. The pharma VP values it because the data is clinical, not administrative. The practice admin values it because the EHR is built for oncology, not adapted from a general system. Same capability. Same differentiation. Different proof below the fold.

The homepage hero addresses neither of these humans. It addresses a board of directors. And it avoids the shared value entirely in favor of mission language ("reimagining the infrastructure") that no buyer from either segment would repeat to a peer.

Two shelves, not one. The pharma buyer shops the real-world evidence shelf (alongside IQVIA, Tempus, ConcertAI, Aetion, TriNetX). The oncology practice buyer shops the oncology EHR shelf (alongside Epic Beacon, Elekta MOSAIQ, Varian ARIA). These two shelves have different buyers, different sales cycles, different evaluation criteria, and different competitive sets.

Flatiron's category language avoids both shelves. The homepage says "healthtech company dedicated to improving cancer care and advancing research." The hero says "reimagining the infrastructure of cancer care." The instinct to avoid naming either shelf is understandable (naming one might alienate the other). But the fix is not to name neither. The fix is to lead with the shared differentiated value that transcends both shelves: Flatiron is the only company that captures oncology data at the point of care. That claim is equally true and equally valuable on both shelves. The current hero says "infrastructure" instead, which sits on no shelf at all.

Real-world evidence in oncology is now in the mid-pragmatist phase. The FDA accepted RWE for the first supplemental oncology indications years ago. The question is no longer "should we use real-world evidence?" It is "whose data is clean enough to submit to the FDA?" That makes the buying decision a data quality evaluation, not a vision-buy.

Oncology EHR is deep pragmatist. Epic Beacon is the safe default. Community oncology practices that switch EHRs are switching for very specific workflow reasons (regimen management, clinical trial matching, prior auth), not because someone "reimagined the infrastructure."

Copy implication: Both buyers need proof language, not visionary language. The pharma VP needs "1,000+ peer-reviewed publications, five years of FDA submissions built on this dataset." The oncology practice admin needs "4,500 providers already run their practice on OncoEMR." Both of those facts exist. Neither reaches the hero.

The chain is intact, and it is unusually strong. The vertical integration (own the EHR, curate the data, serve the evidence to regulators) is a genuine structural advantage no competitor can replicate without building or buying an oncology EHR first. The problem is that this vertical integration thesis appears nowhere in the hero. The hero says "infrastructure" without explaining what it means.

"Flatiron is the only oncology data company that sees the patient at the point of care. Everyone else is reconstructing the story after it happened."

Defensible, structural, and falsifiable. Either Flatiron runs the EHR that generates the data, or it does not. No other RWE competitor (IQVIA, Tempus, ConcertAI, TriNetX) owns an oncology EHR. This sentence does not appear in Flatiron's hero, subhead, or CTAs.

Founded 2012 in New York City by Nat Turner and Zach Weinberg. Both Wharton undergrads. Previously co-founded Invite Media (ad-tech, sold to Google for $81M in 2010 when they were 24). Zero prior healthcare experience. Turner has said publicly that Flatiron was inspired by a family member's cancer diagnosis and the realization that oncology data was siloed, unstructured, and inaccessible to researchers. The founding thesis: what if you could build the data infrastructure that cancer research is missing, starting from the point of care?

• Jan 2013: $8M Series A led by Google Ventures (First Round, LabCorp, SV Angel participating)
• May 2014: $130M Series B led by Google Ventures and First Round
• Jan 2016: $175M Series C led by Roche (Allen & Co, Amgen Ventures, Baillie Gifford participating)
• Apr 2018: Acquired by Roche for $1.9B. Operates as an independent affiliate.
• 2025: Clinical research business divested to Paradigm Health for $78M (narrowing to two pillars: point-of-care and evidence).
• Aug 2024: Reports that Roche is exploring divestiture options with Citigroup (Roche ownership deters competing pharma companies from buying Flatiron data).
• Current CEO: Nathan Hubbard (appointed August 2025, ex-Roche Global Healthcare Ecosystems, ex-Genentech). Replaced Carolyn Starrett (now Strategic Advisor).

1. Roche ownership is the positioning problem no hero can solve. Competing pharma companies (Pfizer, Merck, Novartis, BMS, Lilly) are reluctant to hand patient-level oncology data to a Roche subsidiary. This is why Roche explored divestiture in 2024. The positioning problem is structural, not linguistic. No headline fixes who signs the W-2.
2. Tempus is eating the narrative. Tempus went public in 2024 and owns the "AI plus oncology data" story in trade press and investor coverage. Flatiron's vertical integration thesis (we own the EHR that generates the data) is stronger on the merits, but Tempus controls the category narrative with genomics, sequencing labs, and a public stock ticker.
3. The hero mistakes "don't pick a segment" for "don't say anything specific." A homepage that serves multiple segments does not need to pick one buyer. It needs to lead with the shared differentiated value both buyers recognize. "We capture oncology data at the point of care" works for both the pharma VP and the oncology practice admin. "Reimagining the infrastructure of cancer care" works for neither. The problem is not that Flatiron has two audiences. The problem is that the hero retreats into mission language to avoid choosing, when the right move is to lead with the shared value that makes the choice unnecessary.
4. The clinical research business is gone. Flatiron divested its clinical trials network to Paradigm Health in 2025 for $78M, which removes one of the three original pillars. The remaining two-pillar story (EHR plus evidence) is cleaner but also means Flatiron can no longer pitch the "trial matching at the point of care feeds into the evidence layer" loop. The flywheel lost a gear.
5. The two-shelf positioning trap. Flatiron sits on two shelves simultaneously: oncology real-world evidence (pharma buyer) and oncology EHR (practice buyer). These shelves have different buyers, different competitive sets, different evaluation criteria, and different sales cycles. The homepage tries to avoid choosing by naming neither shelf. The result is a hero that says "infrastructure" — a word that sits on no shelf at all. The fix is not to pick one shelf. The fix is to lead with the shared differentiated value (data captured at the point of care) that is equally true on both shelves, then route the buyer to shelf-specific proof below the fold.
6. The nonconsumer is the largest untapped segment on both shelves. On the evidence side: pharma teams still using CRO-led manual chart abstraction because they trust the proven process and fear the risk of a novel data source in a regulatory filing. These teams are not shopping RWE vendors. They are nonconsumers who have decided the switching risk (FDA rejection) outweighs the switching benefit (speed and cost). Flatiron's 5 years of FDA submissions and 1,000+ publications is the exact evidence that could convert them, but the hero never makes the case. On the EHR side: community oncology practices still running Epic Beacon because switching EHRs is the most painful IT decision in a cancer center. They are nonconsumers who have decided "good enough" beats "purpose-built" given the switching cost. OncoEMR's regimen-based workflow is the argument that could change their calculus, but the hero says "infrastructure" instead of "your EHR doesn't think like an oncologist."

"Reimagining the infrastructure of cancer care" is builder language spoken from the inside of the org chart. Neither buyer (pharma VP or oncology practice admin) describes their own problem as an infrastructure problem. The pharma VP's problem is "my evidence package won't survive FDA review." The oncology practice admin's problem is "my EHR doesn't think about cancer the way my physicians do." Neither of those sentences contains the word "infrastructure."

The navigation is essentially empty. One "Contact us" link. No audience-segmented dropdown. No product pages accessible from the top bar. The entire site architecture asks the reader to self-navigate from two text links in the hero. A pharma VP who arrives from a Google search for "oncology real-world evidence vendor" has no nav cue confirming she is in the right place.

The subhead ("We provide solutions across the industry to learn from and improve every person's experience with cancer") is the most generic sentence in healthcare technology. Every word is a wallpaper word. "Solutions" (what kind?), "across the industry" (which industry?), "learn from and improve" (two verbs when one would do), "every person's experience with cancer" (the mission statement, not the value proposition). A cold reader cannot finish "I need this because ______" with anything specific.

For the homepage hero. The shared differentiated value that both the pharma VP and the oncology practice admin recognize, expressed in language that transcends segment-specific jargon.

• Leads with the shared differentiated value, not a segment-specific pitch. "The only oncology data captured where the treatment decision happened" is equally true and equally valuable to a pharma VP evaluating evidence vendors and a community oncologist evaluating EHRs. Neither buyer is excluded. Neither is pandered to.
• The word Flatiron should own is in the first sentence. "Captured" (vs. reconstructed, vs. abstracted, vs. estimated from billing codes). This is the structural differentiator no competitor can claim without building or buying an oncology EHR.
• Proof stacks in the subhead, not a logo bar. 5M records, 1.5B data points, 1,000+ publications. Three numbers in one sentence, each more credible than the last because peer review is the hardest validation to fake.
• The parenthetical sequence does the competitive work. "Not reconstructed from billing codes. Not abstracted from charts after discharge." Each negation eliminates a class of competitor (IQVIA, CROs, manual abstractors) without naming any of them. The reader fills in the competitor herself.
• Two CTAs route by interest, not by segment. "See the evidence catalog" serves both the pharma VP shopping for data and the oncologist curious about the research her clinic contributes to. "See how OncoEMR works" is the practice-side path. Neither forces the reader to self-identify before clicking.
• Passes the swap test cold. Tempus does not run an oncology EHR. IQVIA does not capture data at the point of care. ConcertAI does not own 1,000+ publications. No competitor can put their logo on this hero.

For a dedicated landing page targeting buyers (across both segments) who are already evaluating oncology data or EHR vendors and comparing Flatiron to the alternatives they know.

• Names the enemy in seven words. "Claims data knows what was billed." Every buyer who has tried to use billing-code data for a clinical question has lived this sentence. The enemy is not a competitor. It is the data format the entire industry has been settling for.
• The punch lands last. "If the patient responded" sits at the end of the sentence where English readers weight it most. This is the exact gap the pharma team cannot tolerate in a regulatory submission.
• Contrast is built into the structure. Billed vs. responded. Claims vs. clinical. Reconstructed vs. captured at the source. The reader understands the value without a second paragraph because the shape of the argument is visible in the syntax.
• The subhead stacks proof in descending order of credibility. 5 million records (scale). 1,000+ publications (peer validation). Five years of FDA and NICE (regulatory precedent). Each number makes the next one more believable.
• The vertical integration wedge finally earns its sentence. "No other evidence vendor owns the data at the source" is the one structural claim Tempus, IQVIA, ConcertAI, and TriNetX cannot make. It passes the swap test cold.
• Two CTAs, both rewards. "See the evidence catalog for your tumor type" gives the VP something to bring to her next team meeting. "Read a regulatory submission case" gives her a precedent, not a pitch. Neither asks her to spend calendar time before she knows the reward.

For a campaign landing page targeting buyers (across both segments) who are not yet shopping for a new data vendor or EHR but whose current approach is quietly underperforming. This variant teaches them something they had not framed yet.

• Opens with the buyer's worst fear, not her wish list. The competitor got to FDA first with real-world evidence. She is still planning a randomized trial that costs $40M and takes three years. The reframe: real-world evidence is no longer supplemental. It is the competitive weapon her rival already deployed. This works equally well for the pharma VP evaluating evidence strategy and the oncology practice admin whose affiliated research program is falling behind.
• Names the specific regulatory context. "Supplemental indication" is the use case where RWE is most accepted and most dangerous to ignore. The buyer who reads this sentence knows exactly which product or program on her portfolio is vulnerable.
• The parenthetical does the differentiation work. "Not reconstructed from billing codes after the fact" eliminates IQVIA, claims databases, and manual chart abstraction in one clause without naming any of them. The reader fills in the competitor herself, which makes the claim more credible than if Flatiron had pointed the finger.
• Closes with competitive pressure, not a feature pitch. "Your competitor is probably already using it" converts better than "our data is high quality" because it activates loss aversion, not aspiration. Buyers do not switch because the new option is good. They switch because falling behind is worse.
• Single CTA, promising precedent. "See which indications Flatiron data has supported" is the thing the buyer was going to try to research anyway. Giving it to her for free earns the next click.

• List segments in the hero. The homepage rewrite leads with the shared differentiated value that transcends segment-specific jargon. Segmentation shows up below the fold (in proof, logos, case studies) and in dedicated campaign landing pages, not in the hero itself.
• Use different positioning for different segments. The differentiated capability (data captured at the point of care) is the same across both buyers. The homepage says it once. Campaign variants express the same positioning tuned for different awareness stages, not different segments.
• Address the Roche ownership problem. This is a strategy problem, not a copy problem. No headline can fix the conflict of interest a competing pharma company feels when buying data from a Roche subsidiary.
• Use any wallpaper word. No "reimagining," "infrastructure," "solutions," "across the industry," "intelligent," "empower," "transform," or "ecosystem."
• Claim anything Flatiron hasn't already published. Every number (5M patients, 1,000+ publications, 22 tumor types, 5 years of FDA/NICE, 1,600 centers) is already on flatiron.com or in a cited press release.
• Write prescriptive strategy. The rewrites demonstrate what the existing substance could sound like if the hero led with shared value and campaign pages handled awareness-stage targeting. They do not tell Flatiron how to solve the Roche ownership problem or the Tempus narrative competition.

Flatiron uses "healthtech company" and "infrastructure of cancer care." Neither is a category a buyer shops. The pharma VP shops "oncology real-world evidence." The practice admin shops "oncology EHR." Both are existing shelves with existing competitors. The homepage avoids naming either shelf in favor of invented category language ("infrastructure") that sits on no shelf at all. A cold reader cannot tell from the hero which budget line Flatiron comes out of.

The hero names no alternative. It does not confront IQVIA's claims data (the pharma buyer's status quo). It does not confront Epic Beacon (the practice buyer's status quo). It does not confront CRO-led manual chart abstraction (the regulatory team's fallback). "Reimagining the infrastructure" gestures at improvement without naming what is being replaced. The status quo is invisible in the copy.

The vertical integration thesis (own the EHR that generates the data, curate it, submit it to regulators) is the single hardest-to-copy attribute in oncology data. No competitor can replicate it without building or buying an oncology EHR first. This thesis does not appear in the hero, the subhead, the CTAs, or the first two scrolls of the homepage. It lives in product sub-pages. Substance exists; viewability time is zero.

The numbers are strong: 5M+ patient records, 1.5B data points, 1,000+ peer-reviewed publications, 22 tumor types, 4 countries, 5+ years of FDA/NICE submissions. These are not vanity metrics. 1,000 publications means 1,000 independent peer-review validations. Five years of regulatory submissions means precedent, not a pilot. None of these numbers appear in the hero. They are buried in product sub-pages and press releases.

Flatiron's two buyers (pharma VP, community oncologist) share a differentiated capability: data captured at the point of care. The homepage does not need to pick one buyer. It needs to lead with that shared value in language that transcends segment-specific jargon, then let segment-specific proof (pharma regulatory cases, oncology workflow testimonials) live below the fold or on dedicated landing pages. The current hero does neither. It retreats into "infrastructure" and forces both buyers to self-sort from text links.

"Reimagining the infrastructure of cancer care" is a thesis statement, not a product promise. A cold pharma VP cannot finish "this does X for me" after reading it. A cold oncology practice admin cannot either. The lingering questions (what does "infrastructure" mean, is this an EHR or a data company, what is the next step after "Contact us") all remain open after the four-second scan. The subhead ("solutions across the industry") makes it worse by adding vagueness on top of abstraction.

OncoEMR, Flatiron Assist (clinical decision support), Precision Medicine (trial matching), OncoAir (mobile), Flatiron Horizon (data engine), Panoramic datasets, regulatory submission services, 5+ years of FDA/NICE collaboration, Caris and Abbott partnerships, global expansion to four countries. The whole product is real and one of the most complete in oncology tech. It is distributed across multiple sub-pages with minimal navigation, so a pragmatist buyer who needs to see the entire package in one scan cannot do so.

Roche acquisition ($1.9B) is itself a trust signal for the evidence side. 1,000+ peer-reviewed publications is the single strongest credibility signal in oncology data (the scientific community has validated the dataset 1,000 times). 20+ top pharma companies as customers. Caris Life Sciences and Abbott as named partners. Google Ventures, a16z, and Roche as investors across the funding history. The credibility is overwhelming. The paradox is that Roche ownership is simultaneously the strongest trust signal (for the oncology practice) and the biggest trust liability (for competing pharma companies).

Five years of direct regulatory submissions to the FDA and NICE is not a compliance checkbox. It is a commercial weapon. Flatiron's data has been used in real regulatory filings that resulted in real approval decisions. This is the most defensible regulatory posture in oncology data, and it is buried in a sub-page where no cold reader will find it. A pharma VP evaluating evidence vendors needs to see "five years of FDA submissions" in the first viewport, not behind two clicks.

Critical distinction: Flatiron is unusual because its regulatory track record functions as a value driver, not just an objection handler. Five years of FDA/NICE submissions creates demand — it is the reason a pharma VP picks up the phone. HIPAA compliance, de-identification methodology, and data governance certifications are objection handlers — they prevent a deal from dying in procurement review but do not create demand on their own. The current site treats both categories with equal emphasis, which buries the commercial weapon (regulatory precedent) alongside the gate-clearing credentials (privacy and de-identification). The regulatory track record should lead. The compliance posture should be easy to find so it clears the gate fast.

The pharma VP says "I need clinical endpoints, not billing codes" and "will this survive FDA review?" The oncology practice admin says "my EHR doesn't think like an oncologist" and "I spend more time fighting the system than treating patients." Flatiron's copy says "reimagining the infrastructure" and "solutions across the industry." The mismatch between buyer voice and vendor voice is the widest gap in this audit. The real sentences are simple, clinical, and specific. The hero is abstract, corporate, and architectural.

Composite read: Flatiron has the widest substance-to-hero gap in this dossier series. The company has a genuinely structural advantage (vertical integration from EHR to regulatory-grade evidence), the largest oncology-specific curated dataset in the world, 1,000+ peer-reviewed validations, and five years of direct regulatory submissions. None of it reaches the hero. Every dimension that underperforms in this audit is a failure of hero execution, not a failure of the product or the strategy. The product is extraordinary. The homepage reads like it was written for a board deck.

The positioning flow: The competitive alternatives are IQVIA claims data and CRO-led chart abstraction (pharma side) and Epic Beacon (practice side). The unique attribute is vertical integration — Flatiron owns the oncology EHR that generates the data and curates it into regulatory-grade evidence. No competitor can replicate this without building or buying an oncology EHR. The value that attribute enables is oncology data with clinical endpoints (treatment response, progression, toxicity) that claims databases cannot reach. The customers who care most are pharma VPs of Real-World Evidence filing supplemental indications where claims data is insufficient, and community oncology practice administrators whose physicians spend hours adapting a general EHR to think about cancer. The market category is oncology real-world evidence (pharma shelf) and oncology EHR (practice shelf). The trend is the shift from "should we use RWE?" to "whose data is clean enough to submit to the FDA?" — a data quality evaluation, not a vision-buy.

Value drivers vs. objection handlers: The value drivers are vertical integration (own the EHR, curate the data), 5M+ patient records across 22 tumor types, 1,000+ peer-reviewed publications, five years of direct FDA/NICE submissions, and OncoEMR's regimen-based workflow. These create demand. The objection handlers are HIPAA compliance, de-identification methodology, data governance certifications, and Roche's financial backing. These prevent deals from stalling in procurement. Uniquely, Flatiron's regulatory track record (FDA/NICE submissions) functions as a value driver, not an objection handler — it is the reason a pharma VP picks up the phone. The site treats all of these with equal emphasis instead of leading with the commercial weapons and clearing the compliance gates second.

The nonconsumption opportunity: On both shelves, the largest addressable segment is nonconsumers. Pharma teams still using CRO-led chart abstraction because they trust the proven process and fear a novel data source in a regulatory filing. Community oncology practices still on Epic Beacon because switching EHRs is the most painful IT decision in a cancer center. Both groups have decided the switching cost outweighs the benefit. Flatiron's substance (5 years of FDA precedent for pharma, regimen-based workflow for practices) is the exact evidence that could convert them. But "reimagining the infrastructure of cancer care" is category-insider language that no nonconsumer would ever click on. The copy that converts a nonconsumer has to start with the problem she already knows she has, not the category she has not yet decided to shop.

• flatiron.com · homepage hero verbatim, nav structure, meta description, category language
• flatiron.com/about-us · founding team, mission statement, company overview
• flatiron.com/real-world-evidence · 5M+ patients, 1.5B data points, 22 tumor types, 4 global networks, FDA/NICE collaboration
• flatiron.com/oncology · OncoEMR, Flatiron Assist, Precision Medicine, OncoAir product descriptions

• Flatiron press release · Jan 2013 $8M Series A (Google Ventures, First Round, LabCorp)
• Crunchbase · May 2014 $130M Series B (Google Ventures, First Round)
• Crunchbase · Jan 2016 $175M Series C led by Roche (Allen & Co, Amgen Ventures participating)
• BioPharma Dive · Feb 2018 Roche acquisition for $1.9B

• Fierce Healthcare · "Roche considering sale of cancer data startup Flatiron Health" (Aug 2024)
• STAT News · Paradigm Health acquires Flatiron clinical research business (Dec 2025)
• FierceBiotech · Flatiron expands hematology datasets (Oct 2025)
• BusinessWire · Flatiron surpasses 1,000 research publications (Oct 2024)
• HLTH Insights · Roche strategic crossroads and Flatiron divestiture analysis (Aug 2024)

• flatiron.com · 4,500+ providers across 1,600+ community cancer centers
• flatiron.com · "20+ top global developers of oncology therapeutics"
• Flatiron press · Caris Life Sciences partnership for multimodal data (Jan 2024)
• Flatiron press · Abbott, Natera integration partnerships
• flatiron.com · University of Colorado Anschutz Cancer Center (research partner)

• iqvia.com · broad claims data, global footprint, incumbent data vendor for pharma
• tempus.com · genomic sequencing plus clinical data, public company (2024 IPO)
• concertai.com · oncology-specific RWE and analytics
• trinetx.com, aetion.com · federated RWE networks
• Epic Beacon · oncology module inside Epic's general EHR (status quo competitor for practice side)
• CB Insights, AEI analysis · Flatiron competitive landscape and oncology data company mapping

• Marker/Medium · Nat Turner and Zach Weinberg founder profile (Wharton, Invite Media, Google acquisition)
• Flatiron press release · Nathan Hubbard CEO appointment (Aug 2025), Carolyn Starrett to Strategic Advisor