What counts as qualifying R&D?

Your work qualifies for R&D tax relief if it is part of a project that seeks an advance in a field of science or technology through the resolution of scientific or technological uncertainty, where a competent professional in that field could not readily resolve the uncertainty. That is the definition in the DSIT Guidelines, and HMRC applies it strictly. The advance must extend the knowledge or capability of the field itself, not just your company.

The definition is narrower than “innovation” and broader than “laboratory research”. Plenty of commercially unremarkable engineering qualifies; plenty of impressive product work does not. This page sets out each element of the test, what falls outside it, and what HMRC expects you to be able to evidence.

The four elements of qualifying R&D

Every claim must show four things:

1. A project. A defined piece of work with an objective, a start and an end. R&D relief is claimed project by project, not on innovation in general.
2. An advance in science or technology. The project must aim to create new knowledge or capability in the field, or appreciably improve what already exists.
3. Scientific or technological uncertainty. At the outset, it must not have been known whether the aim was achievable, or how to achieve it in practice.
4. The competent professional test. The uncertainty must be one that a professional with relevant expertise could not readily resolve using existing knowledge.

The same definition applies whichever scheme you claim under. Which scheme applies to your company depends on your accounting period, size and tax position. Whether your work qualifies does not.

What counts as an advance in science or technology?

An advance means extending the overall knowledge or capability of a field, not just your own company’s. This is the point most weak claims get wrong. If a solution already exists in the field and was simply new to your business, adopting it is not R&D, however much effort the adoption took.

An advance can take several forms:

• creating a new product, process, material, device or service through scientific or technological change
• making an appreciable improvement to an existing one, meaning a genuine technical improvement rather than a cosmetic or stylistic change
• duplicating an existing product or process in a new or appreciably improved way, or where the method behind the original is not publicly known (a competitor’s trade secret, for example)

Two further points matter. Commercial novelty is not technological novelty: a product can be first to its market and involve no qualifying R&D at all, because it assembles established technology in a routine way. And failure does not disqualify a project. If the advance was never achieved, the work of attempting it still qualifies; an unresolved uncertainty is often the clearest evidence that the uncertainty was real.

What is scientific or technological uncertainty?

Uncertainty exists when knowledge of whether something is scientifically possible or technologically feasible, or how to achieve it in practice, is not readily available or readily deducible by a competent professional. If the answer sat in published literature, supplier documentation or standard industry practice, there was no uncertainty, whatever it cost you to find it.

Three distinctions keep claims honest:

• Uncertainty is not the same as difficulty. Work can be complicated, slow and expensive while remaining entirely predictable to an expert. Routine complexity does not qualify.
• Commercial risk alone does not count. Not knowing whether customers will buy the product, whether the budget will hold or whether a competitor will move first is business uncertainty, not technological uncertainty.
• System uncertainty can qualify. Combining components that are each well understood can still involve genuine uncertainty, where the field cannot predict whether or how they will work together as a whole. Routine integration, by contrast, does not qualify.

Who is the competent professional?

The competent professional is someone with the qualifications, knowledge and experience to represent the current state of the art in the specific field of the project. The test runs through their eyes: could such a person readily work out the solution from existing knowledge? If yes, the work is not R&D, however new it felt to the team doing it.

In practice, HMRC expects a claim to be anchored in the judgement of a named individual, usually your own senior technical lead. The Additional Information Form requires the senior internal R&D contact to be named on every claim. When we prepare a claim, we interview the competent professional directly, and their account forms the spine of the technical narrative. A claim reverse-engineered from a cost ledger reads that way to an inspector.

Where does an R&D project start and end?

The R&D project is usually narrower than the commercial project that contains it. It begins when work to resolve the identified uncertainty starts, and it ends when the uncertainty is resolved or the attempt is abandoned. Everything else in the commercial venture sits outside the boundary.

That boundary discipline matters in both directions. Work before the uncertainty is engaged (market research, commercial scoping) and work after it is resolved (marketing, user training, routine maintenance and bug fixing) does not qualify. Activities that directly contribute to resolving the uncertainty, such as designing experiments, building test rigs and analysing results, sit inside it. Drawing boundaries generously inflates a claim and invites an enquiry. Drawing them precisely is what “defensibly prepared” means.

What does not qualify?

Work is not R&D just because it was new to your business, hard, or expensive. The recurring non-qualifiers:

• routine development, configuration and integration using established methods
• cosmetic and aesthetic changes, however commercially valuable
• projects whose only uncertainty was commercial: pricing, demand, funding
• challenges in marketing, management, HR or finance rather than in science or technology
• work in the arts, humanities or social sciences (the definition covers fields of science and technology only)
• straightforward adoption of existing technology, including standard implementation of off-the-shelf software

None of this makes such work worthless. It simply means the relief was not designed for it, and claims built on it are the ones HMRC’s compliance teams take apart.

What does qualifying R&D look like in practice?

Three concept-level sketches, deliberately without client names.

Life sciences. A company developing a purification process for a novel biologic finds that published methods do not predict how the molecule behaves at production scale. Yield collapses, and no literature explains why. The structured experimental work to establish process conditions that preserve yield and stability addresses a textbook uncertainty: the field’s knowledge, not just the company’s, has run out. There is more on how this plays out in our life sciences sector guide.

Software and AI. A team needs a machine learning system to hit accuracy and latency targets that published architectures cannot meet on the data available. Systematic experimentation with model design, training regimes and data handling can qualify. Fine-tuning a well documented model on a new dataset by following the vendor’s playbook does not. We cover the dividing line in when machine learning development qualifies as R&D.

Engineering. A component must meet load, temperature and weight targets in an operating environment for which established materials data does not exist. Standard calculation methods give contradictory predictions, so the team designs a prototype and test programme to establish real behaviour. The iterations, including the failed ones, are the R&D.

What documentation does HMRC expect?

HMRC expects evidence that the uncertainty, the baseline in the field and the systematic work to resolve it were real, ideally recorded while the project ran. HMRC checks roughly one in six R&D claims and has more than 500 staff working on R&D compliance, so assume yours will be read critically. The records that carry weight are the ordinary artefacts of real R&D: design documents, experiment logs, test results, version histories, and notes recording why an approach was abandoned.

Every claim must also be accompanied by the Additional Information Form, which includes project descriptions covering the advance, the uncertainties and the work done. HMRC also offers a non-binding online qualification checker (launched September 2025), which can help a first-time claimant orient themselves, though it does not bind HMRC and is no substitute for a proper eligibility assessment.

If HMRC does open a check into your claim, the quality of this evidence largely decides how it goes. See what to expect from an HMRC R&D enquiry.

Where to go next

If your projects pass this test, the next question is what expenditure you can include: see which costs qualify for R&D tax relief. For the wider picture, all of our R&D tax relief guides are indexed in one place.

Eligibility is rarely a yes or no for a whole company. It is a set of project-by-project judgements, and the honest answer to “do we qualify?” is sometimes “partly”. If you would like a straight answer on your own projects, talk it through with a chartered adviser. A short conversation is usually enough to tell whether a claim is worth pursuing.

Written by Matthew Jones ACA CTA. Last reviewed July 2026.

Sources

• DSIT Guidelines: meaning of R&D for tax purposes — the advance, uncertainty, competent professional and field-scope tests.
• CIRD81910: legal status of the Guidelines — the Guidelines apply to periods beginning after 31 March 2023.
• Check if a project qualifies as R&D — HMRC’s non-binding online qualification checker.