The two-phase development approach separates feasibility R&D (Phase 1) from MDR-compliant development (Phase 2). Phase 1 answers "does this work?" without the full weight of Annex II documentation. Phase 2 begins only after feasibility is proven, and produces the technical file, risk file, and QMS records required for CE marking. Startups that skip Phase 1 burn through runway documenting products that never should have been built. Startups that skip Phase 2 ship devices that cannot be certified.

By Tibor Zechmeister and Felix Lenhard. Last updated 10 April 2026.

TL;DR

  • Documenting every prototype, every iteration, and every internal test to MDR-compliance standards from day one slows development by roughly 90 percent in the companies we have seen it happen.
  • Phase 1 is feasibility. Fast, experimental, cheap. No obligation to produce MDR-compliant records because the device does not yet exist in a form that will be placed on the market.
  • Phase 2 begins when you know what you are building. It produces the records required by MDR Article 10 and Annex II — technical documentation, QMS records, risk file — for the specific device that will be certified.
  • The transition point between phases is a formal design-freeze decision. Before it: speed matters most. After it: traceability matters most.
  • The two-phase approach is not a loophole. It is how disciplined MedTech companies actually ship.

Why this matters for your startup

There is a Graz-based company we have worked alongside that decided, on day one, to document everything to full MDR compliance. Every prototype iteration. Every bench test. Every internal meeting. Every design change. They wanted to be "audit-ready from day zero" — the phrase you hear at every regulatory conference.

The result: development slowed by roughly 90 percent. The product that could have been on the market in 2020 is still not on the market in 2026. Six years of over-documentation paralysis, for a device whose fundamental design changed five times during that period. Every one of those changes dragged a compliance trail behind it. Most of the documentation they produced was thrown away because the product it described no longer existed.

Then there is a Salzburg-based SaaS startup. Their algorithms were unproven. Before touching a single MDR document, they ran a Phase 1: collected retrospective hospital data, validated the core technology, confirmed the algorithms worked on the actual patient population they were targeting. Only after Phase 1 gave them confidence did they move to Phase 2 — MDR-compliant development, QMS build, technical file, the whole machine. They saved years of wasted documentation on a product that might never have worked.

Both companies were run by smart people. One of them is dying. The difference is whether they understood the two-phase approach.

What "two-phase" actually means

The two-phase development approach is the deliberate separation of medical device development into two sequential phases with different rules, different tempo, and different documentation standards.

Phase 1 — Feasibility. Research and development in the ordinary sense. You are figuring out whether the technology works, whether the clinical concept is sound, whether the user flow makes sense, whether the device can be manufactured at a reasonable cost. The output of Phase 1 is knowledge, not a regulated product. No device is placed on any market. No device is used on patients outside the framework of a proper clinical investigation or research ethics approval.

Phase 2 — MDR-compliant development. The work required by Regulation (EU) 2017/745 to take a specific, frozen design through conformity assessment and onto the market. This is where Article 10 of the MDR lives — the manufacturer's general obligations, including the quality management system requirement in Article 10(9), the technical documentation requirement referencing Annex II and Annex III, and the post-market surveillance system obligation. This is where the risk management file per EN ISO 14971:2019+A11:2021 gets built. This is where the clinical evaluation per MDR Article 61 is executed.

The distinction matters because MDR obligations attach to devices intended to be placed on the market. A feasibility prototype built to answer "does the signal processing algorithm work at all?" is not a device being placed on the market. A frozen design entering verification and validation is.

The instinct to document everything in MDR format from the beginning comes from fear. Auditors like documentation. Investors like "audit-ready." Regulatory consultants like billable hours. All three create pressure to start the compliance machinery on day one. And all three are wrong for startups whose core technical risk has not been resolved yet.

The MDR text — what the regulation actually requires

MDR Article 10 sets out the general obligations of manufacturers. Article 10(9) of Regulation (EU) 2017/745 requires manufacturers of devices (other than investigational devices) to establish, document, implement, maintain, keep up to date, and continually improve a quality management system that ensures compliance with the Regulation in the most effective manner, proportionate to the risk class and type of device.

The phrase "other than investigational devices" matters. Devices in genuine research and development — not yet intended for placing on the market — sit under a different part of the Regulation. Investigational devices used in clinical investigations are governed by Articles 62 to 82 and Annex XV, not by the full manufacturer obligations in Article 10.

Annex II of the MDR specifies the technical documentation required for a device that is going through conformity assessment — the device description, design and manufacturing information, the GSPR checklist, the benefit-risk analysis, the risk management file, and so on. Annex III specifies the post-market surveillance technical documentation.

Nothing in the MDR requires you to produce Annex II documentation for a prototype that will never be placed on the market in that form. The obligation attaches to the device you intend to certify, not to every sketch, experiment, or bench test that led to it.

This is the legal basis for Phase 1. You are not cheating. You are correctly identifying the moment at which MDR obligations begin.

The transition point — design freeze

The transition from Phase 1 to Phase 2 is not a gradient. It is a decision. At some point, the design of the device is frozen — this specific architecture, these specific components, this specific intended purpose, this specific user population — and from that moment forward the work to be done is conformity assessment work, not exploration work.

A competent design freeze decision is based on several things happening simultaneously. Feasibility is demonstrated. The intended purpose is stable and written down. The classification is understood. The rough conformity assessment route is known. The risk analysis has identified the major hazards and you have at least a conceptual approach to controlling them. The clinical evaluation strategy is at least sketched.

Before the design freeze, you are iterating fast, changing things, failing cheaply. After the design freeze, every change goes through change control, every document gets version-controlled in the QMS, and every modification triggers the right subset of re-verification and re-validation activities.

Founders who never make the design freeze decision stay in Phase 1 forever. They iterate indefinitely. The product never stabilizes enough to certify. This is failure by another name.

Founders who make the design freeze decision too early commit to a design that does not work. Phase 2 is expensive and slow. Every wasted month in Phase 2 on the wrong design is ten times more costly than a wasted month in Phase 1.

The judgment call on when to freeze is one of the highest-leverage decisions a MedTech founder makes. It is also one of the hardest to automate, which is why having a sparring partner who has been through it before — with multiple devices, across multiple classifications — is worth more than almost any other early-stage regulatory investment.

What each phase should actually contain

Phase 1 — Feasibility work. Rapid prototyping. Benchtop experiments. Retrospective data analysis where available. Algorithm validation on existing datasets. User interviews with actual clinicians in the target specialty. Workflow observation in real hospital or care environments. Basic technical risk analysis (informal — not yet the full ISO 14971 file). Intended purpose drafts. Classification sketches. Conversations with potential Notified Bodies about scope. Early conversations with early clinical partners. The Felix rule applies throughout: strip everything that does not move the needle. If a feature does not clearly advance feasibility, cut it. You will rebuild it in Phase 2 if it turns out to matter.

Phase 2 — MDR-compliant development. Full QMS build aligned with EN ISO 13485:2016+A11:2021. Risk management file per EN ISO 14971:2019+A11:2021, constructed in parallel with design. Software development lifecycle per EN 62304:2006+A1:2015 if software is involved. Usability engineering per EN 62366-1:2015+A1:2020. Biocompatibility per EN ISO 10993-1:2025 for any patient-contact material. Electrical safety per EN 60601-1 if applicable. Clinical evaluation per MDR Article 61 and Annex XIV. Technical documentation per Annex II. PMS system per Article 83 and Annex III. And the whole document set reviewed and audited internally before it ever touches a Notified Body.

Note what is not in Phase 1: no QMS. No technical file. No formal risk management file. No clinical evaluation plan in its final form. Just fast, focused exploration of whether the device concept is worth certifying at all.

Note what is not in Phase 2: no more architectural reinvention. No fundamental changes to intended purpose. No adding new indications halfway through. Phase 2 is execution, not exploration.

Common mistakes startups make with the two-phase approach

  • Starting Phase 2 too early, because the founder confuses "we have a working prototype" with "we have a frozen design." A prototype that has never been used outside the founder's lab is not frozen. It is a hopeful sketch.
  • Never leaving Phase 1, because the founder loves building and hates documentation. Phase 1 becomes a permanent hobby. Runway runs out before any regulated product exists.
  • Treating Phase 1 as an excuse to skip ethics. Feasibility does not mean testing on employees, family members, or friendly doctors without proper research ethics approval. The red line between research and unauthorized clinical use is absolute. We have seen founders cross it. The consequences — regulatory, legal, personal — are not survivable.
  • Forgetting that Phase 1 still generates information that Phase 2 will need. Keep decent records during Phase 1. Not QMS records, but lab notebooks, design decisions, test results, user interview notes. In Phase 2 you will need to reconstruct the rationale for design choices, and you will not remember.
  • Documenting Phase 1 work in MDR format anyway, "just in case." This is the Graz over-documentation pattern. It looks diligent. It is actually paralysis dressed in compliance clothing.

The Subtract to Ship approach to phasing

The minimum viable regulatory strategy begins with asking what must genuinely be done to take this specific device, for this specific intended purpose, through conformity assessment — and cutting everything else. The two-phase approach is that principle applied to the time axis.

Phase 1 subtracts every activity that does not answer a feasibility question. Phase 2 subtracts every activity that does not contribute to the technical documentation, risk file, clinical evaluation, or QMS records that the Notified Body will actually look at.

The transition is where the Subtract to Ship discipline earns its keep. A founder with a good regulatory partner and a clear intended purpose can move from Phase 1 to Phase 2 in weeks. A founder without those things spends months in the transition and never notices they have already left feasibility behind.

There is a story from the consumer side that illustrates the pattern. During the pandemic, one of Felix's portfolio companies — Vulpine — faced a forced subtraction. Capital tightened. Lead times collapsed. The team had to decide what actually mattered and ship only that. Three things became clear in the space of a few weeks: side features nobody used, scarcity as a creative forcing function, and a founder's capacity for stubborn focus when the alternative was dying. Nothing about that story is MedTech, but everything about it is the two-phase mindset. You do not get to do everything. You get to do the work that survives contact with reality, and you do it under constraints.

Reality Check — Where do you stand?

  1. Can you state, in one sentence, whether you are currently in Phase 1 or Phase 2? If the answer is "both at the same time," you are in neither.
  2. Is your intended purpose stable enough to write down in a single paragraph that has not changed in 30 days?
  3. Have you identified your classification with confidence, or are you still hoping it might be Class I?
  4. If a Notified Body walked into your office tomorrow and asked to see your QMS records, would you show them real QMS records or a folder of hopes?
  5. If the answer to question 4 is "real records" — are those records genuinely needed because you are in Phase 2, or did you start the QMS machinery too early?
  6. How much of your runway have you spent on documentation for prototypes that no longer exist?
  7. Who, on your team or in your advisory circle, is qualified to make the design freeze call? If the answer is "nobody," you have a gap to close before Phase 2.

Frequently Asked Questions

Is the two-phase approach legal under the MDR? Yes. MDR Article 10(9) applies to manufacturers of devices placed on the market, not to research and development activities on prototypes that are not yet placed on any market. Phase 1 is ordinary pre-market R&D. Phase 2 is where the full manufacturer obligations begin. The distinction is explicit in the Regulation's structure.

Can I test my Phase 1 prototype on patients? Only within the framework of a properly approved clinical investigation under MDR Articles 62 to 82 and Annex XV, or under an appropriate research ethics framework that permits the use. Informal testing on employees, family members, or friendly clinicians without ethics approval is not a shortcut — it is a crime in most EU jurisdictions and a hard red line in this book and in Tibor's practice.

When exactly should I freeze the design? When three conditions are simultaneously true. First, feasibility is demonstrated with real data, not just optimism. Second, the intended purpose has stopped changing. Third, you have a clear enough understanding of classification and conformity assessment route that you can estimate Phase 2 cost and time. Freezing earlier is premature. Freezing later is Phase 1 forever.

Do I need a QMS during Phase 1? Not the full ISO 13485 QMS. But you should keep disciplined records — lab notebooks, design decisions, test results, user interviews, key emails. These become input to the Phase 2 technical documentation and risk file. Losing them wastes months in Phase 2 reconstructing what you already knew.

How do I explain Phase 1 to an investor who expects "MDR-ready from day one"? By teaching them what MDR-ready actually means. An investor who thinks every prototype must be Annex II-compliant is an investor who has funded a company that is burning their money on paperwork. The right answer is: "Phase 1 proves the technology works. Phase 2 turns the working technology into a certified medical device. We are in Phase [X] and the design freeze is expected by [date]." Any competent MedTech investor understands this. Investors who do not are a warning sign.

Sources

  1. Regulation (EU) 2017/745 of the European Parliament and of the Council of 5 April 2017 on medical devices, Article 10 (general obligations of manufacturers), Articles 62–82 (clinical investigations), Article 83 (post-market surveillance system), Annex II (technical documentation), Annex III (technical documentation on post-market surveillance), Annex XV (clinical investigations). Official Journal L 117, 5.5.2017.
  2. EN ISO 13485:2016 + A11:2021 — Medical devices — Quality management systems — Requirements for regulatory purposes.
  3. EN ISO 14971:2019 + A11:2021 — Medical devices — Application of risk management to medical devices.
  4. EN 62304:2006 + A1:2015 — Medical device software — Software life-cycle processes.
  5. EN 62366-1:2015 + A1:2020 — Medical devices — Part 1: Application of usability engineering to medical devices.
  6. EN ISO 10993-1:2025 — Biological evaluation of medical devices — Part 1: Requirements and general principles for the evaluation of biological safety within a risk management process.
  7. EN 60601-1:2006 + A1 + A12 + A2 + A13:2024 — Medical electrical equipment — Part 1: General requirements for basic safety and essential performance.

This post is part of the MDR Fundamentals & Regulatory Strategy series in the Subtract to Ship: MDR blog. Authored by Felix Lenhard and Tibor Zechmeister. If you are standing at the Phase 1 / Phase 2 boundary and need a second opinion on whether your design is actually ready to freeze, Zechmeister Strategic Solutions works with founders on exactly that decision.