The guidelines on the meaning of R&D are the same for every sector: an advance in science or technology, sought by resolving scientific or technological uncertainty that a competent professional could not readily resolve. What differs is where that work hides and what HMRC most often pushes back on. This page walks through the main sectors; the industry guides go deeper on each.
Where qualifying work sits by sector
| Sector | Usually qualifies | Usually does not | Watch for |
|---|---|---|---|
| Software | New architectures, algorithms, performance and scalability work where the approach was uncertain | Configuration, routine integration, use of established frameworks, UI design | HMRC's software guidance and the competent professional test |
| Manufacturing and engineering | New or improved products, materials and processes, prototypes, trial runs to resolve uncertainty | Routine production, quality control, tooling with known outcomes | When the uncertainty is resolved; consumables in products sold |
| Life sciences | Discovery, preclinical and clinical development, formulation, process scale-up | Post-approval marketing studies, routine release testing | Overseas trials; contracted out work; grants |
| Construction and engineering services | Novel structural, geotechnical, materials or systems solutions with real uncertainty | Standard designs, compliance with codes, project management | Who decided to undertake the R&D under the contract |
| Food and drink | New formulations, shelf-life and process development | Taste, recipe variation and packaging aesthetics | The aesthetic exclusion |
| Energy and cleantech | Materials, storage, generation and grid technology development | Installation and commissioning of proven systems | Grant funding and the treatment of subsidised expenditure |
Software
Software is the largest category of UK claims and the one HMRC has scrutinised hardest. The qualifying work is where the team did not know whether an approach would work, and a competent software professional could not readily have known: a new data structure to meet a latency requirement, a model that had to reach a required accuracy on the company's data, a distributed system that had to remain consistent under failure. HMRC's guidance is explicit that using established tools and frameworks in the way they are intended, integrating systems through documented interfaces, and producing a new product by applying known techniques are not R&D, however commercially novel the product. The advance must be in the field of computer science or software engineering, not in the company's own product set.
Agile teams should treat the backlog, the ticket history and the commit log as the evidence of the work to resolve uncertainty, with tickets tagged to the project and the uncertainty they address. The R&D ends when the uncertainty is resolved, which in software is usually well before release.
Manufacturing and engineering
Manufacturers under-claim because the R&D happens on the shop floor and nobody calls it research. A new product is obvious. Process development is not: a new parameter set for welding an unfamiliar alloy, a fixture that had to be designed and tested to hold a tolerance, a trial run to establish whether a line could handle a new material at rate. Each is R&D if it resolves a scientific or technological uncertainty, and the materials consumed in trials are consumable items. The guidelines are clear that a prototype built to test the design is part of the R&D, while a prototype or first production unit built for sale is not, and that the cost of consumables incorporated in products sold is excluded.
The line to watch is the end of the R&D. Once the process runs in production, improvements with known outcomes are routine engineering, and claims that run every project to the year end invite enquiry.
Life sciences
Pharmaceutical, biotechnology and medical device development is close to the paradigm of R&D, and the difficulties are structural rather than scientific. Clinical trials run through contract research organisations are contracted out R&D, so from April 2024 only trials carried out in the UK generally qualify, with an exception for work that must be done abroad for reasons such as patient populations or regulatory requirements. Payments to clinical trial volunteers are a qualifying category in their own right. Many life sciences companies are loss-making and R&D intensive, so ERIS, with its 186% deduction and payable credit of up to 14.5% of the surrenderable loss, is often the better route; the ERIS page sets out the intensity test.
Other sectors
Construction and engineering services. Qualifying work exists in novel solutions, but much of it is done under contract. Since April 2024 the company that made the decision to undertake the R&D claims; a contractor doing work to a client's specification generally cannot, while a contractor that chose to develop a new method to deliver the contract may be able to. The contract and the records decide.
Food and drink. Formulation to achieve a technical result, such as shelf life, texture or stability, qualifies; recipe variation for taste and packaging design do not, because aesthetic and commercial advances are outside the guidelines.
Energy and cleantech. Development qualifies in the ordinary way. Grant-funded projects can be claimed under the merged scheme, because the old rules reducing relief for subsidised expenditure no longer apply for accounting periods beginning on or after 1 April 2024.
Financial services and other service businesses. Claims rest on software, and the software guidance applies with full force.
Illustrative example
Illustrative example. Three hypothetical companies, one project each. Mersey Analytics rebuilds its search index to cut query latency from two seconds to under 200 milliseconds, evaluating four indexing strategies against a benchmark; its lead engineer records why none of the documented approaches met the target. R&D, with the ticket history as evidence. Pennine Castings develops a casting process for a new aluminium alloy, scrapping 40 trial pours before porosity was controlled. R&D, with the scrap as consumables. Severn Foods reformulates a cereal bar to taste more like caramel. Not R&D: taste. Severn's parallel project to keep the bar from going stale for 12 months in a new film is a different matter.
Related pages
Official sources
Educational content only, not tax advice. See the disclaimer.