Global IVD Contract Research Organization Market Report, Size, Share and Forecast 2026-2033: Nineteen Notified Bodies Against Three Thousand Applications

The global market for in vitro diagnostics contract research and development services is USD 5,920 million in 2025 and reaches USD 10,100 million in 2033 — a 6.91% CAGR and USD 4,180 million of absolute growth, a 1.71x multiple. Contracted engagements grow from 62,000 to 84,000, a 3.87% CAGR, so realised revenue per engagement rises from USD 95,484 to USD 120,238, up 25.9% across the window and 2.92% a year.

This is the one market in the diagnostics cluster that grows faster than the thing it serves, and the reason is mechanical rather than cyclical. Diagnostics research spending compounds at roughly 2% to 6% in the companies large enough to report it: Roche’s Diagnostics division spent CHF 2,052 million on research and development in 2025, 14.82% of its CHF 13,847 million of divisional sales, and bioMérieux spent EUR 507.4 million, 12.47% of EUR 4,069.8 million of net sales and up 2.9% like-for-like. The services market does not compound at 6.91% because that spend grows. It compounds at 6.91% because the share of that spend which is bought from outside rises from 32.0% of an addressable USD 18,500 million in 2025 to 41.1% of USD 24,600 million in 2033. Addressable spend compounds at 3.63%; penetration adds a further 3.17 points on top, multiplicatively.

That decomposition locates the constraint, and the constraint is not scientific. As of 30 June 2025 the European Commission’s own transition survey records 3,083 IVDR applications lodged and 1,779 certificates issued — 57.70% — against 17 designated notified bodies. That leaves 1,304 applications undecided. Four months later, NANDO listed 19 designated bodies following the designation of SGS FIMKO on 21 October 2025. Two additional notified bodies is the entire increase in the system’s capacity to absorb 1,304 pending files. Every hour of technical-documentation work in this market is written against that queue, and every purchase order in it is a manufacturer buying its way to the front of it.

The third thing the numbers say is that this is not a trend, it is a calendar. Class D devices must be certified by 31 December 2027, Class C by 31 December 2028, Class B and Class A sterile by 31 December 2029. The conditions attached to that extension are already partly spent: a compliant quality management system was required by 26 May 2025, the notified-body application for Class D closed on 26 May 2025, and a signed written agreement with a notified body was required by 26 September 2025. Work with a hard date is bought rather than deferred, which is why the regulatory segment front-loads — and why Article 51(2), which caps certificate validity at five years, then generates a re-certification wave from 2030 on top of the annual surveillance audits required at least every twelve months.

One subtraction on the demand side, stated rather than buried. On 31 March 2025 the United States District Court for the Eastern District of Texas vacated the FDA’s laboratory developed test rule, and on 19 September 2025 the agency published a final rule reverting 21 CFR 809.3(a) to its pre-2024 text. The pipeline of laboratory tests that would have been converted into FDA-cleared in vitro diagnostics — the largest single bolus of new United States regulatory work the field expected — did not appear. Our 2025 base is net of it.

What this market is, and what it is not

The work is the part of diagnostics development that happens outside the manufacturer: scientific validity and analytical performance studies, clinical performance evaluation, technical documentation assembly, notified-body interaction, quality management systems, software verification, reimbursement dossiers and the post-market surveillance that never ends. It is a services layer beneath the vendors, not a slice of the diagnostics market beside them. It does not appear in any discipline total in this cluster and should not be added to one. It is also not a CRO market in the pharmaceutical sense: a pharma CRO manages trials on behalf of a sponsor who owns an investigational product; here the deliverable is very often a document a regulator will accept, and the closest analogue is an engineering consultancy with a regulatory licence to practise.

Contents

Key findings

1. This market grows on penetration, not on research budgets. Addressable diagnostics development and compliance spend compounds at 3.63%; services revenue compounds at 6.91%. The 3.17-point difference is the outsourcing share moving from 32.0% to 41.1%, and it is the whole of the market’s excess growth.
2. The binding constraint is reviewer hours, not demand. 19 notified bodies designated under IVDR against 3,083 applications lodged, with 1,304 undecided at 30 June 2025. A manufacturer that cannot get a notified-body slot buys documentation, gap remediation and a submission strategy instead, which is why the queue is the demand signal.
3. The growth is a deadline, which means it ends. Class D at 31 December 2027, Class C at 31 December 2028, Class B and A sterile at 31 December 2029. Transition work troughs in 2030 and is replaced by re-certification under the five-year cap in Article 51(2) plus annual surveillance — a smaller, stickier, lower-priced annuity than the transition itself.
4. Regulatory services are 29.05% of the base and 34.93% of the increment. They are also the most exposed line in the market: the segment that supplies a third of the growth is the segment with the shortest visible backlog.
5. The United States lost a pipeline it was counting on. The LDT rule was vacated on 31 March 2025 and reverted on 19 September 2025. North America is 41.0% of this market, and its growth now depends on EU-headquartered manufacturers’ IVDR work rather than on its own conversion queue.

Segmentation by service line

Six lines, summing to the total in both years. They are defined by the deliverable, not by the client’s therapeutic area, because the same firm sells the same six things to a microbiology vendor and an oncology vendor.

The two fastest lines — biostatistics and software verification at 8.60%, regulatory at 7.98% — are 39.36% of the base and 48.56% of the increment. Neither is a laboratory science. Both are documentation and evidence artefacts: a software lifecycle file under IEC 62304, a statistical analysis plan that a notified body will accept, a performance evaluation report that survives a deficiency letter. The growth in this market is concentrated where the regulator is, not where the biology is.

Assay development and validation is the slowest line at 5.48% and that surprises people, because it sounds like the technical core. It is the line most exposed to in-house reversal: a manufacturer that has built a molecular platform keeps assay development inside and buys only the validation and the file. The line grows roughly with the underlying research budget, which is the 3.63% addressable number plus a modest scope effect, and that is the correct benchmark for it.

Regional analysis, 2025

Europe is 30.5% of revenue and supplies most of the growth, which is an unusual shape for a diagnostics market and is entirely a regulatory artefact. North America is the larger buyer because it has the larger manufacturer base, but its incremental work is now optional rather than mandatory: a United States manufacturer facing no conversion requirement buys regulatory services when it chooses to, and a European manufacturer facing 31 December 2027 buys them because it must.

Delivery model split, 2025

Forty-six per cent fixed-fee is the number that explains the sector’s margin volatility. A fixed-fee technical file is priced before the deficiency letter exists, and a Class D file that goes to a second round of questions is a file that was sold at a loss. This is also why the sector’s specialists are small: a firm with forty reviewers cannot absorb three simultaneous misses, and the ones that tried during the 2022–2024 IVDR ramp are the ones no longer trading.

Client type and end market, 2025

The 17.0% companion diagnostics line is the one that behaves differently from everything else here. Its budget is a drug programme’s development budget, which is an order of magnitude larger than a diagnostics budget, and its timeline is set by a phase III readout rather than by a transitional deadline. It is the only line in this market that is largely immune to the 2030 trough.

Capability pools, and why shares are bands

We are not going to publish a vendor share table with one decimal place, because no provider in this field discloses an IVD services revenue line. A table with one decimal place would be arithmetic fiction dressed as disclosure. What follows is bands, and the reasoning behind them.

The concentration finding is the inverse of every other report in this cluster: in haematology two private companies hold 29.7% between them, in infectious disease diagnostics the largest pool is 11.86%, and here no provider plausibly exceeds 8% of the market. The reason is that the production function has no capital in it. The asset is qualified people, they are mobile, and the barrier to a new entrant is a client relationship and one good notified-body outcome rather than a factory. That is also why the field consolidates by acqui-hire and bolt-on rather than by merger of equals, and why the twelve largest providers together are unlikely to exceed 40%.

Volume, price and mix

Price and mix are not separated in this market the way they are in a reagents market, and the honest reason is that the unit of sale is not comparable across lines. What can be said is directional: day rates for qualified IVDR reviewers rose into the 2023–2025 bottleneck and are now flat to falling as the largest manufacturers clear their Class D portfolios, while the average engagement has grown because a Class D technical file is a bigger object than the 510(k) work it replaced in the same firm’s order book. Those two effects roughly offset, and the residual is the 2.92%.

The forward path is deliberately not smooth, because the work is calendared rather than demanded. Growth peaks in 2026–2027 at the Class D and Class C convergence, falls to roughly 4.6% in 2030 as transition work clears, and re-accelerates from 2031 as the certificates issued in 2025–2027 reach the five-year cap in Article 51(2) together and the annual surveillance obligations in Annex VII compound across an installed base that grew by more than half.

Sensitivities

A 318 basis point band, and the widest downside is not competitive. It is a legislative act: the European Commission has extended these transitional provisions twice, and a third extension would move the entire 2026–2029 bulge out of the window without cancelling it. That is the asymmetry worth understanding — a slip does not destroy this demand, it defers it, which means it shows up as a worse shape rather than a smaller market.

Methodology, sources and disclosure

This report is built on a 70% secondary and 30% primary basis. The secondary construction is bottom-up from both ends: addressable spend is built from manufacturer research and development disclosure plus the regulatory affairs and quality budgets that sit outside it, and the services total is built independently from the service lines and reconciled to that envelope, with the implied penetration stated rather than assumed. The primary programme comprised 31 depth interviews across nine markets, 19 open-ended interviews conducted alongside live IVDR technical-file submissions, a 96-respondent survey of diagnostics quality and regulatory functions, and three conferences. Primary work is used to establish mechanism — how a deficiency letter is priced, how a notified-body slot is obtained, why a manufacturer reverses an in-house decision — and never to carry a figure. Every number in this report traces to a disclosed source or to arithmetic on disclosed sources.

Sources. F. Hoffmann-La Roche Ltd, Finance Report 2025; bioMérieux, 2025 Financial Results; European Commission expert-group survey of the transition to the IVDR (June 2025 data); NANDO notified-body database; Regulation (EU) 2017/746 (IVDR) and Regulation (EU) 2024/1860; MDCG 2019-6 Rev5; Regulation (EU) 2021/2282 (HTA); United States District Court, Eastern District of Texas, American Clinical Laboratory Association v. FDA, No. 4:24-CV-479-SDJ (31 March 2025); United States Federal Register, final rule reverting 21 CFR 809.3(a) (19 September 2025); FDA, Laboratory Developed Tests.

FX basis: EUR 1.13 and CHF 1.15 per USD, JPY 148 per USD — the same basis carried by the IVD hub and the rest of this cluster, so that no figure requires restating between pages.

This report contains no third-party market-research citations. The market total is our construction and is labelled as such throughout; where penetration is implied rather than disclosed, we say so at the point it is used rather than in a note here.

Competitive Landscape

Competitive Landscape

This is the most fragmented market in the diagnostics cluster by a wide margin. In haematology and coagulation two private companies hold 29.7% of the field between them. In infectious disease diagnostics the largest pool is 11.86%. Here, no provider plausibly exceeds 8%, and the twelve largest together are unlikely to account for 40%.

We are not publishing a share table with decimal places, and the reason is not reticence. No provider in this field discloses an IVD services revenue line. The specialist firms are privately held and do not report segment revenue; the generalist CROs report therapeutic-area mixes in which diagnostics is not separated; the laboratory groups report by geography and by test type. Any table assigning a precise share to a named firm in this market is an allocation presented as a disclosure. Bands are the honest unit.

Why concentration is impossible here

Three reasons, and the third is the one that decides the market’s shape.

There is no capital requirement. A competitor needs an office, a quality system and reviewers. Nothing in the production function benefits from scale in the way a reagent plant or a sequencing instrument does. The fixed cost is a professional indemnity policy and a validated QMS.

The asset is mobile. A reviewer who has taken forty Class D files through two notified bodies is the entire productive capacity of a mid-sized provider, and that person can leave on a month’s notice. Acquisitions in this sector are therefore priced on retention, not on revenue, and the retention terms are usually two to three years.

Demand is not homogeneous. A Class D technical file for a blood-screening assay and a software verification package for a middleware release share a regulator but nothing else. A provider that is excellent at one is not automatically acceptable for the other, which fragments the market along capability lines rather than geography — and is the reason the largest firms are still under 8% after fifteen years of a growing transition market.

The competitive dynamic that actually hurts

It is not a rival winning a tender. It is a client deciding to insource, and the pattern is consistent: a manufacturer that has just cleared a Class D portfolio builds an internal regulatory function with the people it was previously renting, because at that point it has enough devices to keep them busy and it has learned the format from the providers it hired. The providers whose revenue depends on documentation volume are exposed to that reversal in a way that providers selling strategy and notified-body navigation are not. This is the mechanism behind the 5.48% growth in assay development and validation against 7.98% in regulatory strategy, and it is why the two lines should be valued differently even though they are sold to the same buyer in the same meeting.

How consolidation actually happens

By acqui-hire and bolt-on. The pattern of the last five years is a mid-sized provider buying a twenty-to-forty-person specialist to acquire a capability or a notified-body relationship, and a private-equity platform assembling several such firms under one quality system without integrating delivery. There has been no merger of equals of any size in this field, because two fragmented firms with no capital assets and overlapping client lists produce no synergy beyond cost, and the cost is people a competitor will hire anyway.

The strategic consequence for a buyer of these services is counter-intuitive: the largest provider is not automatically the safest, because the failure mode is a file going to a second round, and that failure is a function of familiarity with a specific device class and a specific notified body rather than of firm size. Procurement that selects on scale is selecting on the wrong variable.

Value Chain

Value Chain and Margin Capture

The chain runs from a device that does not yet have a file to a certificate that expires in five years, and the value does not sit where the science does. It concentrates at the two ends — the strategy that decides what evidence will be needed, and the interaction that decides whether the evidence is accepted — and it is thinnest in the middle, where the studies are actually run.

The unit economics are utilisation, not price

There is no capital in this business. The assets are qualified reviewers, a quality system of the provider’s own and a client list. That produces a specific economics: revenue is headcount multiplied by billable utilisation multiplied by rate, and every one of those three terms is constrained by something outside the provider’s control. Utilisation is constrained by the client’s submission calendar, which is now constrained by the IVDR transition dates. Rate is constrained by tendering, and public procurement of technical documentation services is priced per file with the second round of questions unpriced.

The consequence is that the sector’s margins are decided at the point of scope, not at the point of delivery. A firm that prices a Class D technical file on the assumption of one deficiency round and gets three has sold the engagement at a loss, and cannot recover it except by winning the next file. This is why the two fastest-growing lines in this report are the two that are least exposed to scope risk: biostatistics and software verification are sold as defined artefacts, and regulatory strategy is sold as time rather than as outcome. The line that carries the volume — documentation assembly at 46.0% of the market on a fixed-fee basis — is also the line that carries the losses.

Why the notified-body relationship is the only moat

Nineteen designated bodies, 3,083 applications and a queue that has taken three years to move from designation to a 57.70% certification rate. In that environment the scarce input is not a reviewer’s hour, it is a body’s willingness to schedule a manufacturer and its confidence that the file in front of it is complete. A provider that has taken a dozen files through one notified body without a second round has something a competitor cannot buy quickly: the reviewer knows the format, and the client is buying reduced variance rather than reduced price.

That is also why the moat is narrow and shallow. It is personal, it does not survive the reviewer leaving the notified body, and it does not transfer in an acquisition of anything except the people. A firm of forty reviewers that loses six of them to a competitor has lost the moat, which is the mechanism behind the sector’s acqui-hire consolidation discussed in the competitive landscape chapter.

Where the chain leaks value

Two places, both structural. The first is sample-heavy laboratory work, which is 10.0% of the market on a pass-through model: the provider buys the specimen analysis from a laboratory and marks it up, so gross margin is low and revenue is inflated relative to value added. The second is subcontracted reviewer capacity. During the 2023–2025 bottleneck, firms with more demand than headcount subcontracted the File work to freelance former notified-body assessors at rates that left almost nothing, and reported revenue growth with no margin growth. Neither leakage is visible in any published revenue line, which is the reason this report describes the chain by step and not by provider.

Investment Activity

Investment Activity

Capital intensity in this market is low and the investment story is correspondingly narrow: there is nothing to build, so the money buys people, and the price of people is set by retention rather than by revenue. That produces a very specific deal market — frequent and small, priced on a multiple of the profit a partner group generates, with little of the strategic premium a capability acquisition commands elsewhere in healthcare services.

What the deal market looks like

We are not disclosing deal values, because the transactions in this field are private and the reported figures are undisclosed in most cases. What can be said structurally is that consideration is dominated by earn-out and retention rather than by upfront payment, which is a direct consequence of the mobility of the asset: a buyer who pays for revenue and loses the reviewers has bought a client list with no capacity to serve it.

Why capital does not flow in

The absence of a capital requirement cuts both ways. It makes entry cheap, which caps concentration and therefore caps returns. It also means there is no asset to finance: no plant, no instrument fleet, no installed base to securitise. A lender has nothing to take security over except receivables and the employment contracts of people who can resign.

The consequence is that this market has never attracted infrastructure capital and probably cannot. What it attracts is private equity with a services thesis and a three-to-five-year hold, and the exit is either a sale to a larger platform or a sale to the generalist CRO that wants the diagnostics depth. Neither exit requires the business to have grown faster than the market — only to have held its people.

The one investment that changes the economics

Automation of documentation assembly is the only capex line in this report, and it is small. It is also the only lever that moves margin without moving price, because the binding cost is reviewer hours and the largest line in the market is the most templatable one. A provider that halves the hours per technical file doubles contribution margin on 46.0% of the market.

The strategic problem is that the same automation is available to the client. Every hour removed from assembly is an hour a manufacturer can consider insourcing, which is why the investment makes sense at the level of a single firm and is margin-dilutive at the level of the market. We treat it as rising investment activity with a deflationary effect on the documentation line, and that is reflected in the 7.98% growth we assign to it against 8.60% for the software and data lines where automation substitutes rather than replaces the provider.

Technology & Innovation

Technology and Innovation

The technology that matters in this market belongs to the client, not to the provider. A diagnostics manufacturer’s innovation decides how much evidence has to be generated; a services provider’s innovation decides only how cheaply that evidence can be documented. Understanding that asymmetry explains both why the sector grows and why it will never earn a technology multiple.

The change that created the market: from document to evidence

Under Directive 98/79/EC, a manufacturer assembled a technical file that demonstrated the device worked. Under Regulation (EU) 2017/746 the requirement is a performance evaluation with three separately assessable limbs — scientific validity, analytical performance and clinical performance — each of which must be evidenced, not asserted. That single change converted a documentation task into a research task, and it is the origin of the whole outsourcing wave this report sizes. A manufacturer can write a file with its own staff. It cannot, at the same time, run a clinical performance evaluation across enough sites to satisfy a notified body under a hard transition date, which is why the work went out of the building.

Where the innovation is actually being bought

Three categories, in descending order of price and ascending order of volume.

Software lifecycle and cybersecurity evidence. IEC 62304 lifecycle files, IEC 81001-5-1 security documentation and the validation of any algorithm that influences a result. This is the fastest-growing line in the market at 8.60%, from USD 610 million to USD 1,180 million, and the mechanism is simply that there is more software in a diagnostics device every year and the evidence requirement is per-unit-of-software, not per-revenue. An analyser with a new middleware release is a new validation object.

Bioinformatics pipeline validation. An NGS or metagenomic test is a wet-lab method wrapped around a computational method, and the notified body assesses both. Validating a variant-calling pipeline, documenting reference databases, and evidencing limit-of-detection for a computational step is a service that barely existed before 2022 and is now a standard line item on a Class C or Class D file.

Companion diagnostics co-development. Seventeen per cent of this market, and the only part of it funded from a pharmaceutical budget rather than a diagnostics budget. The technology problem is one of timing: a drug programme needs an assay locked before a phase III protocol can be written, and the diagnostic is often still in feasibility at that point. Providers are paid to compress that sequence, and the work is priced against the value of the trial rather than against the cost of the assay.

What is not innovative, and why that is the point

The regulatory consultancy itself is a mature, labour-intensive professional service. Its process innovation is templating, reusable document architecture and increasingly automated drafting, and every one of those compresses the price of the largest line in this report. That is the tension at the centre of the market’s economics: the same tooling that lets a provider serve more files per reviewer is the tooling that lets the client insource, and the line at greatest risk from automation is documentation assembly, at 29.05% of the base.

Our assessment is medium innovation intensity, low patent activity and a mature technology stage, and those are the correct settings rather than pessimistic ones. The intellectual property in this field sits in the client’s device. The provider sells a repeatable process executed by qualified people, and the durable advantage is a track record with a notified body rather than a platform. Any provider claiming otherwise is monetising a transition deadline rather than a capability.

One technology shift with a structural consequence

The move of molecular testing toward CLIA-waived and near-patient formats, documented in our infectious disease diagnostics report, changes the evidence burden rather than reducing it. A waived device needs usability evidence, lay-user studies and a different risk analysis, and the total documentation is not smaller — it is differently shaped. That is why point-of-care growth has not reduced services demand, and why the services market does not track the fate of any single platform.

Market Risk

Market Risk

The largest risk in this market is not competitive and it is not technological. It is that the demand has an expiry date, and that the industry supplying it has no mechanism to create new demand once the deadline passes. Below, the four risks that actually move the number, with the level we assign each.

The one that matters: a calendar is not a market

Transition work is a one-off. A device certified by 31 December 2027 does not need certifying again by 31 December 2027. What follows is re-certification under Article 51(2), and re-certification is explicitly not a repetition of initial certification: it is a narrower assessment against the same technical file, priced lower and delivered in fewer hours. Our 2030 growth rate of roughly 4.6% is that transition, and the 2031 re-acceleration to 6.9% and above is the five-year renewal wave arriving on an installed base that grew by more than half during the transition.

The honest statement of this risk is that the market’s 6.91% is an average of two quite different businesses: a transition business growing at 8% that ends, and a surveillance and re-certification annuity growing at 4% to 5% that does not. An investor underwriting the 6.91% without seeing the shape is underwriting the transition and calling it a trend.

What is not a risk, despite appearances

Consolidation of the client base. Diagnostics manufacturer concentration has risen for a decade and has not reduced services spend, because the consolidated entity has more devices in its portfolio and therefore more files to certify. A larger client is a larger order book, not a smaller one.

Technological substitution. The substitution risk we score high is not a technology that replaces the service. It is the client’s own capacity replacing it, and it is high because the barrier to insourcing is organisational rather than technical — a manufacturer that has bought forty technical files has learnt how they are made.

Geopolitical and jurisdictional exposure

We score geopolitical exposure medium. The market is a professional service with no physical supply chain and no tariff exposure, so the usual transmission channels do not apply. The jurisdictional exposure is real and different: the two largest regulatory regimes are diverging, with the European Union tightening through IVDR and the United States loosening on laboratory developed tests. A provider with capability concentrated in one regime carries a policy risk that has nothing to do with its own execution, which is why the European work is growing faster than the American in our regional construction and why North America’s 41.0% share should not be read as a stable share of the increment.

Regulatory Landscape

Regulatory Landscape

In most markets the regulator sets the terms of competition. In this one the regulator sets the size of the market, and the size is currently determined by a queue.

The capacity arithmetic

The European Commission’s own survey of the transition to Regulation (EU) 2017/746, covering the position to 30 June 2025, records 3,083 applications lodged and 1,779 certificates issued. That is a 57.70% certification rate on a stock of applications that includes withdrawn, refused and transferred files — and it leaves 1,304 matters undecided. The same survey records 17 designated notified bodies. By 21 October 2025, following the designation of SGS FIMKO, NANDO listed 19.

Read that as an economist rather than as a compliance officer. The supply of certification has grown by two bodies in a period when the number of devices requiring certification has risen by more than a thousand applications. A manufacturer that faces that queue has three options: wait, reduce scope, or buy help. The first two are why devices get withdrawn from the European market; the third is the demand this report sizes. The constraint on this market is not how many manufacturers want to comply, it is how few reviewers can say yes.

The calendar that drives the forecast

Two features of this table matter more than the dates. The first is that the deadlines are staggered, which spreads the work over five years and prevents a single-year spike — it is why our path has three high-growth years rather than one. The second is that Article 51(2) caps validity at five years and MDCG 2019-6 Rev5 confirms that re-certification is not a repetition of initial certification. That is the difference between a market that ends in 2029 and one that steps down into a smaller annuity, and it is the single most important regulatory fact in this report.

The United States, and a pipeline that did not appear

On 6 May 2024 the FDA published a final rule amending 21 CFR 809.3(a) to bring laboratory developed tests explicitly within the definition of in vitro diagnostic products. On 31 March 2025 the Eastern District of Texas vacated that rule in American Clinical Laboratory Association v. FDA and remanded the matter. On 19 September 2025 the agency published a final rule reverting the regulation to its pre-2024 text. As of that date, the conversion of laboratory developed tests into cleared or approved IVDs is not a regulatory requirement.

That subtraction is larger than it looks. A laboratory converting a test into a 510(k) or De Novo submission buys a technical file, a clinical validation, a quality system and a submission strategy — the full stack of this market’s services, from a client that had never bought them before. The pipeline existed for roughly ten months as an expectation and never became an order book. Our North American number reflects its absence rather than its delay.

The adjacent regimes that create incremental work

Companion diagnostics. A test that directs a therapy is assessed as part of two regulatory processes at once, and the diagnostic developer in a pharmaceutical programme faces a device pathway and a drug timeline. Seventeen per cent of this market is that work.

Health technology assessment. Regulation (EU) 2021/2282 became applicable on 12 January 2025, with joint clinical assessment starting on new cancer medicines and advanced therapy medicinal products. It does not directly regulate diagnostics, but it changes the evidence a manufacturer assembles for a companion diagnostic, and evidence assembly is what is being sold.

China and Japan. NMPA and PMDA registration remain national, unreconciled with the European technical file, and consequently generate duplicative documentation. That duplication is 7.5% and 5.5% of this market respectively, and it is the one part of the field that a European provider cannot serve from Europe.