Real-World Evidence Platform Selection: A Buyer’s Guide

The 2026 RWE Platform Market Landscape

Three structural shifts have redrawn the RWE platform map in the eighteen months leading into 2026, and every buyer needs to understand them before starting a vendor evaluation. Ignoring any of these three shifts leads to the same predictable outcome: a platform selected on twelve-month criteria that stops fitting the organization’s actual needs somewhere between month eighteen and month thirty. Buyers who treat the market as static and evaluate today’s data assets in isolation from vendor ownership, regulatory acceptance, and data model direction consistently end up rebuilding programs at exactly the moment those programs should be delivering evidence to support submission timelines.

The first shift is regulatory legitimization. FDA’s Real-World Evidence Program, established under Section 3022 of the 21st Century Cures Act, has now supported label expansions, external control arms in rare disease and oncology, and post-marketing safety commitments across dozens of approved products.² In December 2025, FDA updated its guidance to allow sponsors to submit RWE for medical device applications without patient identification data. That is a signal that the agency is comfortable with tokenized, de-identified data linkage as the underlying architecture for regulatory-grade evidence.³ On the European side, the EMA’s Data Analysis and Real World Interrogation Network (DARWIN EU) now accesses data from approximately 250 million patients across the EU regulatory network, with 108 research topics assessed and 88 studies completed or ongoing as of the most recent annual report.⁴ RWE is no longer a “nice to have” appendix; it is a first-class evidence type for both agencies.

The second shift is consolidation. Roche’s acquisitions of Flatiron Health (2018) and TriNetX (2024) placed two of the largest oncology and federated data assets under one parent. Datavant’s May 2025 announcement of its acquisition of Aetion combined the leading tokenization and connectivity platform with the leading regulatory-grade RWE analytics platform, creating what the combined companies describe as “an end-to-end RWE platform” for life-sciences research.⁵ These are not neutral events for buyers. They compress vendor choice, they change roadmap priorities, and they introduce data-access risk when a competitor of your therapeutic area now sits under the same corporate parent as your data provider.

The third shift is the maturation of common data models. The Observational Medical Outcomes Partnership (OMOP) Common Data Model, developed and maintained by the OHDSI community, has become the de facto standard for regulatory-grade observational research. Regulators including FDA and EMA now recognize OMOP-mapped data as a foundation for reproducible, transparent RWE, and major pharmaceutical companies such as Novartis and AstraZeneca have internally mapped key datasets to OMOP for both post-approval and late-phase research.⁶ Any platform that cannot deliver data in OMOP (or credibly demonstrate a path to it) has become difficult to defend for regulatory use cases.

Five Platform Archetypes and How They Differ

Before evaluating individual vendors, it helps to categorize them. Every RWE platform in the market today falls into one of five archetypes, and each archetype answers a different research question. Choosing the wrong archetype is more expensive than choosing the wrong vendor within the right archetype.

Vendor Profiles: The Eight Platforms Sponsors Actually Shortlist

The following profiles cover the eight platforms that appear most consistently on senior life-sciences leaders’ shortlists in 2026. Each profile is short by design; the goal is to describe what the platform is genuinely strong at, where it struggles, and which use cases it fits.

Datavant-Aetion

The combined entity, announced in May 2025, brings together Datavant’s tokenization and connectivity infrastructure with Aetion’s regulatory-grade analytics environment. Aetion Activate, launched in 2024, expanded the software portfolio to serve both biostatisticians and no-code users with a collaborative, auditable environment for evidence generation.⁵ The strength here is the combination: tokenized linkage across data sources plus a fit-for-purpose analytics layer with strong provenance. The risk is that integration between the two product lines is still in progress, and the combined roadmap is not yet fully public. Best fit: sponsors who need external control arms, safety commitments, or FDA-facing evidence that must link a trial cohort to downstream real-world outcomes.

Flatiron Health (Roche)

Flatiron remains the strongest oncology-specific dataset in the US, built on structured plus abstracted EHR data from more than 280 community and academic oncology practices. Its regulatory-grade external control arm work has supported multiple FDA label expansions. Since Roche’s 2018 acquisition, Flatiron has operated with a partial firewall from its parent, but sponsors who compete directly with Roche in specific tumor types should evaluate that relationship carefully. Best fit: oncology-focused sponsors, particularly for label expansion, external control arms, and post-marketing commitments in solid tumors.

TriNetX (Roche)

TriNetX operates the world’s largest federated real-world data network, connecting more than 240 healthcare organizations and 13,000+ clinical sites across 20+ countries.⁸ The federated architecture means patient data never leaves the healthcare organizations that generate it; queries run against member sites and return aggregated results. As of late 2024, the Global Collaborative Network included approximately 153.5 million EHRs, the US Collaborative Network approximately 117.2 million, with additional networks for LATAM, EMEA, and APAC.⁸ Best fit: global feasibility, site identification, cohort discovery, and multi-country observational studies. Weaker fit when you need patient-level record linkage or curated unstructured data at scale.

IQVIA

IQVIA offers the broadest integrated set of RWD assets, technology (including OMOP CDM tooling), and regulatory services in the industry, with commercial and clinical data covering hundreds of millions of lives globally.⁹ The strength is scale plus a global services organization that can execute end-to-end RWE studies. The weakness is that IQVIA is simultaneously a CRO, a data provider, a technology platform, and a consultancy. Depending on the deal structure, sponsors sometimes end up buying more than they intended. Best fit: global sponsors that need one-stop delivery for a portfolio of RWE studies, especially where local regulatory affairs support is required.

Optum

Optum Labs Data Warehouse covers 300+ million de-identified lives combining commercial claims, Medicare fee-for-service, EHR-derived data, and socioeconomic determinants, spanning more than 20,000 mapped clinical variables and over 1 billion prescriptions from 150+ US payers.¹⁰ Optum’s NLP capabilities on clinical notes are among the most mature in the market. Best fit: HEOR, market access, disease burden, and US-centric outcomes research. Weaker fit when the research question requires non-US data or highly curated tumor-level detail.

ConcertAI

ConcertAI orchestrates real-world data, clinical expertise, and applied AI for oncology and other complex diseases, with data assets used in more than 500 peer-reviewed publications. Its Patient360 asset combines longitudinal EHR data, curated clinical detail, and molecular data across multiple community and academic sites. Best fit: oncology sponsors that want an independent alternative to Flatiron (particularly those competing with Roche in specific tumor types), and studies that need clinico-molecular depth.

PicnicHealth

PicnicHealth takes a fundamentally different approach: patients themselves consent to have their medical records collected from every provider they have seen, and PicnicHealth combines machine learning with clinician review to create structured, longitudinal datasets. The company has entered the oncology data market and is developing cohorts across breast, colorectal, bladder, and prostate cancers, and Roche/Genentech has signed a strategic data partnership with PicnicHealth.¹¹ Best fit: rare disease, patient-consented registries, and studies where standard EHR extraction cannot deliver the completeness required.

Verantos

Verantos generates what it calls “high-validity” RWE by combining structured data with clinical narrative extraction and AI, with an explicit focus on measured accuracy, completeness, and traceability at the record level.¹² The platform positions itself as regulatory-grade by design, not by retrofit. Best fit: sponsors preparing evidence for label expansion, indication broadening, or safety commitments where FDA reviewers will scrutinize source-record fidelity.

Honorable mentions worth watching

Beyond the eight above, several platforms show up on shortlists for specific use cases. Truveta operates a health-system-owned data collaborative with more than 30 major US health systems contributing normalized EHR data; sponsors focused on US oncology and cardiovascular research often shortlist it as an EHR-primary alternative to Flatiron. Tempus combines molecular sequencing data with longitudinal clinical data and is the natural choice when a research question requires biomarker-linked outcomes. Komodo Health remains a strong claims-primary option with 330 million de-identified patient journeys and a differentiated conversational AI layer that opens exploratory analytics to non-technical users. HealthVerity operates as a semi-open marketplace and identity manager where sponsors can construct custom linkages across multiple data suppliers. Veeva Crossix and Clarify Health address patient-journey and DTC attribution use cases that sit closer to commercial than to regulatory. Any of these can be the right answer for a specific question; none of them replace one of the top eight for a broad-based RWE program.

The Nine Evaluation Criteria That Matter

Most RWE platform RFPs we see contain 40 to 80 evaluation criteria. That is too many. In practice, nine criteria drive the decision, and everything else is downstream of these.

1. Data coverage and depth for your therapeutic area

Ask for specific patient counts in your indication, not global lives. A platform with 300 million lives and 4,000 patients meeting your inclusion criteria is worse than a platform with 50 million lives and 40,000 matching patients. Request a feasibility count against your protocol before you shortlist.

2. Standardized common data model

OMOP CDM has emerged as the dominant regulatory-grade model, with strong FDA and EMA support and demonstrated cross-database interoperability.⁶ PCORnet CDM remains relevant for US comparative effectiveness research. Sentinel Common Data Model is required for FDA Sentinel work. Ask what CDMs the platform delivers natively and what mapping is required to move to a different one.

3. Analytics capabilities and statistical validity

Can biostatisticians use the platform to run pre-specified protocols with locked analysis plans and full audit trails? Can they extend it with custom code? What causal inference methods are supported natively (propensity score matching, target trial emulation, marginal structural models)? A platform that only supports drag-and-drop analyses will not survive FDA scrutiny on a label expansion.

4. Regulatory-grade evidence generation

Has the platform actually been used in an FDA or EMA submission that received a positive decision? Ask for named submissions (with sponsor permission), not marketing claims. The FDA Framework for the RWE Program explicitly evaluates fitness of both data and study design. Vendors that cannot describe their approach to both dimensions are not regulatory-grade.²

5. Integration flexibility and interoperability

Can the platform ingest your internal clinical trial data? Can it link to a tokenized cohort from Datavant or a competing linkage provider? Can outputs be exported to your enterprise data lake? A closed platform that trades performance for portability will make your next migration painful.

6. IRB and consent handling

RWE research typically requires either individual authorization or an IRB waiver under HIPAA.¹³ Ask how the platform documents IRB oversight, whether it maintains master authorization frameworks with data suppliers, and how it handles state-level privacy laws that go beyond HIPAA (California CMIA, Texas HB 300, Washington My Health My Data).

7. Data provenance and quality documentation

For every field you plan to use in a regulatory analysis, the vendor should be able to describe the source, the transformation, the quality checks, and the known limitations. If they cannot, that field is not regulatory-grade. Ask to see a data quality report at the variable level, not the dataset level.

8. Vendor stability and roadmap

Given the consolidation wave, ask directly: are you in active acquisition discussions? Who is your parent company today, and who might it be in eighteen months? What is the roadmap for the specific product I am buying, not the corporate product line?

9. Contract terms that protect optionality

Request data extract rights on termination, IP rights for derived cohorts, audit rights for regulatory inspections, and clear language on how new data assets acquired by the vendor (or its parent) will be made available to existing customers. Sponsors that fail to negotiate these terms end up locked in.

The criteria most sponsors under-weight

Two criteria on this list almost always receive too little weight in early-stage vendor conversations: consent handling (criterion six) and contract optionality (criterion nine). Both feel like legal problems that can be solved late. In practice, both determine whether the platform is usable at all. A platform that cannot produce clean IRB documentation for a rare disease cohort will not be defensible in a submission review. A contract that lacks data extract rights turns a vendor swap into a two-year rebuild. The pattern we see across successful RWE programs is that the general counsel, chief privacy officer, and regulatory head are involved in vendor selection from week one, not brought in at the contract-signature stage to sign off on decisions already made.

A Weighted Vendor Evaluation Matrix

The following matrix provides a starting weighting for the nine criteria above. Weights should be adjusted to reflect your primary use case. A sponsor building an external control arm for a rare disease submission should weight regulatory-grade evidence generation and data depth higher than integration flexibility. A commercial analytics team should weight coverage and analytics accessibility above regulatory-grade capability.

Pricing Models and Total Cost of Ownership

The RWE platform market has settled into four pricing models, each with implications for how the platform will actually be used inside your organization. Industry analysis identifies subscription pricing as the dominant model, with pay-per-usage growing quickly among smaller sponsors and license-based models used for enterprise-wide deployments.¹

Model 1: Subscription

Annual or multi-year subscription with a defined user population and data access scope. Range: typically $200,000 to $3 million per year for mid-cap pharma, higher for global players with multiple therapeutic areas. Predictable, easy to budget, but can create a “use it or lose it” dynamic where teams justify the subscription by running low-value studies.

Model 2: Pay-per-Study or Pay-per-Usage

Per-study or per-query pricing, sometimes with a small platform access fee. Range: $50,000 to $500,000 per study depending on scope. Best for sponsors running fewer, high-value studies; risky when a study balloons in scope mid-analysis and pricing was not capped.

Model 3: Enterprise License

Full platform access with unlimited use across the organization, typically for global pharma with a mature RWE function. Range: $3 million to $15 million+ annually. Delivers the lowest per-use cost when the organization uses the platform heavily, but requires a mature internal function to realize that value.

Model 4: Milestone / Value-Based

Fixed-fee delivery tied to defined milestones (e.g., delivery of a regulatory-grade study report). Provides budget certainty and shifts execution risk to the vendor; less common with data-network platforms, more common with services-heavy delivery.

Which model fits which sponsor

Emerging biotechs with one or two lead assets and a small internal RWE function should almost always start with pay-per-study or milestone pricing. The commitment is bounded, the deliverable is defined, and the sponsor keeps optionality as its pipeline evolves. Mid-cap pharma with three to eight active RWE workstreams typically finds subscription pricing most efficient, provided the subscription is scoped to the therapeutic areas actually in use rather than to global data access the organization will not exercise. Global pharma with a mature RWE function running dozens of studies per year almost always gravitates to enterprise licenses, but should structure them with annual true-up clauses tied to actual usage so the pricing tracks value delivered rather than shelf-ware. Across all four models, the negotiation lever that matters most is not per-record pricing; it is data extract rights, IP terms, and the ability to walk away at contract renewal without losing the derived cohorts your team has built.

Six Implementation Traps That Sink RWE Programs

Across dozens of RWE platform implementations, six failure modes recur. They are all avoidable and all preventable in the first 90 days if named early.

A 90-Day Proof-of-Concept Framework

The single most useful step in RWE platform selection is a structured proof of concept before signing a multi-year contract. Proofs of concept reduce uncertainty in the software selection process by giving teams the opportunity to see the tool in action on their terms, identify limitations early, assess usability, and validate whether the platform fits workflows and expectations.¹⁴ The framework below is designed to fit inside 90 days and cost no more than $75,000 to $150,000 per vendor.

What to do if a POC is not politically feasible

Occasionally a sponsor’s leadership decides a POC is too slow, too expensive, or too politically fraught, and asks the team to move directly to a full contract. When that happens, the risk-reduction tools that most closely approximate POC discipline are (a) requiring the vendor to include a first study delivery inside the initial contract term with clear acceptance criteria, (b) structuring the first-year commitment as a limited scope with expansion options rather than an enterprise deal, (c) inserting a termination-for-convenience clause with a short notice period during the first six months, and (d) requiring named references from three sponsors currently running the same use case you plan to run. None of these fully replaces a real POC, but together they reduce the cost of an incorrect vendor choice from “multi-year, multi-million-dollar rebuild” to “an inconvenient year and a difficult conversation with procurement.” The best sponsors we work with treat this as a fallback rather than a default; if leadership pressure to skip the POC is intense, the underlying question is usually about internal capability rather than about the vendor, and that deserves its own conversation.

Conclusion

Selecting a real-world evidence platform in 2026 is a portfolio decision, not a single-vendor decision. The most sophisticated life-sciences organizations are running two or three complementary platforms wired together with tokenization and a common data model, negotiating contracts that preserve optionality as the market consolidates, and investing in the internal epidemiology and biostatistics capability that turns platform access into regulatory-grade evidence. The organizations that struggle are the ones that treat platform selection as a procurement exercise or a therapeutic area team’s preference, and that discover the limitations of their choice only when a submission deadline is 90 days out.

The market context matters here. Consolidation is accelerating, common data models are converging on OMOP, tokenization is becoming table stakes, and both FDA and EMA are integrating RWE more deeply into regulatory decision-making with each successive guidance cycle.¹⁵ A platform that looks acceptable today may be an acquisition target next quarter, and a contract signed without extract rights or roadmap protections may look very different when the parent company sits under a competitor’s balance sheet. None of this is a reason to defer the decision. It is a reason to make the decision with better questions, tighter proof-of-concept discipline, and clearer language on what happens when the market shifts underneath the contract.

Sakara Digital works with pharma and biotech organizations building real-world evidence programs that need to hold up under FDA and EMA scrutiny. If you are evaluating RWE platforms, negotiating a multi-year contract with a data vendor, or trying to sequence a proof-of-concept before you commit, we are happy to have an independent, non-vendor conversation about where to start and what to insist on. The right answer is rarely the biggest logo; it is the platform that answers the specific evidence question that matters most to your next milestone, at a total cost of ownership your leadership will actually recognize, on a contract your legal team will be able to unwind when the next inevitable acquisition reshapes the market again.