CIR technical file: the 2026 audit-proof template

Quick answer. France's research tax credit (CIR) covers 30 % of eligible R&D expenditure up to 100 million euros. The declared figure is not enough: it is the technical justification file, which demonstrates R&D within the meaning of the OECD Frascati Manual, that secures the credit during an audit.

Many business owners think the CIR is decided in the tax return and form 2069-A-SD. In reality, disputes happen elsewhere: in the ability to prove, operation by operation, that the work qualifies as research and development. The tax authority almost never challenges the arithmetic. It challenges eligibility. And eligibility is demonstrated in a single document: the technical file.

This guide sets out the template we use to build a solid technical file, the pitfalls that trigger a reassessment, and what changed with the 2025 Finance Act, applicable to expenditure incurred from 15 February 2025.

The CIR rests on article 244 quater B of the French General Tax Code. The calculation looks simple: 30 % of eligible research expenditure up to 100 million euros, then 5 % beyond (50 % in the overseas departments up to 100 million euros). But this rate applies only to expenditure attached to genuinely eligible R&D operations.

The technical file is the document that establishes this eligibility. It is not a commercial summary of the project or a marketing brochure. It is a scientific and technical demonstration, structured operation by operation, showing that the work meets the criteria for research.

The tax authority (DGFiP) may, during an audit, request a scientific assessment from the ministry in charge of research (MESR). It is this expert, not a tax inspector, who decides whether the work constitutes R&D. Your technical file is therefore written to be read by a scientist in your field.

Our reading. A poorly documented CIR is not a risky CIR: it is a CIR that has not yet been justified. As long as the technical file does not exist, the declared credit is a fragile provision, not a settled gain.

R&D status is assessed against the OECD Frascati Manual, the reference adopted by administrative doctrine. Five criteria structure the analysis. A project must meet them cumulatively to be eligible.

The most discriminating criterion in practice is scientific or technical uncertainty. A development difficulty solved by applying known techniques is not R&D. There must be a barrier that accessible knowledge did not allow to be lifted directly.

What the authority looks at. The MESR expert first looks for the state of the art and the technical barriers. If these two elements are missing or vague, the operation is reclassified as routine development, even if the engineering work was real and costly.

A technical file is built operation by operation. The MESR released a model operation description sheet, updated in 2024, with one sheet per operation of limited length, in the order of ten pages. This structure has become the implicit reference of audits.

Each section carries a verifiable statement. The state of the art is not a decorative bibliography: it serves to prove the absence of an existing solution. The barriers are not scheduling constraints: they are knowledge obstacles.

How to build the file, step by step#

1. List the R&D operations of the financial year, distinct from the commercial projects that carry them.
2. For each operation, write the state of the art by reviewing publications, patents and market solutions as at the date of the work.
3. State explicitly the scientific or technical barriers encountered, distinguishing them from project management difficulties.
4. Describe the experimental approach, keeping the trace of hypotheses, trials and failures.
5. Present the results, whether positive or negative: a documented failure remains R&D.
6. Link the declared basis to the work described, justifying the time spent by each assigned person.
7. Have the sheet reviewed by the scientific lead before finalising the figures.

In practice. The technical file is written as the work proceeds, not in the year of the audit. Teams that wait three years to reconstruct a state of the art produce a dated, incomplete document that is fragile in front of the expert.

The 2025 Finance Act tightened several CIR parameters, for expenditure incurred from 15 February 2025. The 2026 Finance Act maintains the scheme without major reform: these 2025 rules remain the working basis. The general framework and the detail of eligible expenditure are in our CIR 2026 guide.

Abolishing the young doctor status changes the hiring trade-off for several tech SMEs. The doubling of the staff cost basis and the enhanced flat rate no longer apply: a recently hired doctorate holder no longer brings the same extra credit as before.

2026 points of vigilance. Three mistakes have recurred since these rules took effect: still applying the 43 % flat rate, including patent or watch costs that are now ineligible in the basis, and keeping a young doctor calculation for hires made after 15 February 2025. Each one weakens the declared credit.

In the files we take over after a first audit, the grounds for challenge look alike. They rarely concern the calculation, almost always the justification.

• State of the art absent, or reduced to a list of references without demonstrating the insufficiency of existing solutions.
• Scientific barriers confused with scheduling, budget or integration difficulties.
• Technical file written in commercial language, with no scientific content the expert can use.
• Time spent not tracked: no timesheet linking staff to the R&D operations.
• Public subsidies not deducted from the basis, when they must be.
• Outdated parameters kept (43 % flat rate, patent costs, young doctor) after 15 February 2025.
• Confusion between the CIR and the innovation tax credit (CII), which follows distinct rules.

The underestimated risk. The costliest risk is not the rejection of an isolated expense, but the reclassification of a whole operation as routine development. When the expert concludes there is no R&D, the entire credit attached to the operation falls, not just a few percent of the basis.

A software-publishing SME had declared a substantial CIR over three financial years, relying on its engineers' time and a functional description of the product. The technical work was real. But the file contained neither a state of the art nor a statement of the barriers: it described what the software did, not why obtaining it was scientifically uncertain.

During the audit, the MESR expert concluded that the work was advanced software engineering, but not R&D within the meaning of Frascati, for lack of an identifiable barrier that accessible knowledge could not lift. Reconstructing the state of the art three years later did not reverse the analysis. The lesson is consistent: a technical file is documented during the work, not after the assessment notice.

R&D subcontracting. Expenditure entrusted to providers follows a specific regime. The technical file must identify the subcontracted operations, distinguish what qualifies as R&D from a mere technical service, and keep the supporting documents from the chosen body. The traceability of the eligible operation also applies to the outsourced share.

Multi-year project. A project spread over several financial years is split into annual operations. Each year requires its own justification: the state of the art evolves, some barriers are lifted, others appear. A file frozen in year one does not cover year three.

SME requesting immediate refund. The CIR is offset against corporate income tax; the surplus is refundable, and immediately so for SMEs within the European meaning. It is a real cash-flow asset, but a refund request draws attention. It is better to file a complete technical file with the request than to build it under the pressure of an audit.

Trade-off. Documenting finely costs R&D time. Many teams hesitate between declaring broad and fast, or declaring narrow and substantiated. Our position is constant: a more modest but fully justified CIR secures cash flow better than a maximalist CIR exposed to reclassification. The right scope is the one the technical file can support.

• The CIR covers 30 % of eligible R&D expenditure up to 100 million euros, but the credit holds only through the technical justification file.
• Eligibility is assessed against the Frascati Manual criteria: novelty, creativity, uncertainty, systematic approach, transferability.
• The state of the art and the statement of barriers are the two items the MESR expert looks for first.
• Since 15 February 2025, the operating flat rate is 40 % of staff costs, patent and watch costs are no longer eligible, and the young doctor status is abolished.
• Public subsidies are deducted from the basis of eligible expenditure.
• A technical file is written during the work, operation by operation, not in the year of the audit.