Does engineering or manufacturing work qualify for R&D tax relief?

This note covers engineering contractors, fabricators, subcontractors and precision manufacturers, toolmakers, process and production engineers, and product manufacturers.

Where the R&D usually is

Product development is the obvious half — designing something to meet a requirement no existing product meets, where it was not clear at the outset that it could be done or how.

Process development is the half that gets missed, and in manufacturing it is often the larger one. Making an existing product to a tolerance the current process cannot hold, at a yield it cannot reach, from a material it cannot handle, or at a rate it cannot sustain, is capable of being an advance in its own right. So is developing a joining, forming, coating or inspection method where no established technique works on the geometry or material in front of you. The product may not change at all.

The work also qualifies where it fails. What matters is the intention to achieve an advance, not whether the uncertainty is resolved — a process route that is developed, trialled and abandoned because it cannot be made to work is still R&D up to the point the work stops.

Two things do not disqualify you. That the advance is not visible in the finished article does not matter; the test is capability in the field, not what a customer sees. And where a competitor has plausibly solved the same problem but keeps the method to itself, the work can still be an advance.

Where it usually is not

• Making things to an established method, however tight the tolerance and however skilled the work. Production is production.
• Routine adaptation of an existing design to a new size, material or customer, where a competent engineer would know how to do it.
• Selecting between existing equipment, materials or suppliers, and commissioning plant to the manufacturer’s instructions.
• Quality control, inspection and rework in the ordinary course. Testing that feeds back into resolving an uncertainty is different from testing that confirms a part meets its drawing.
• Cosmetic and aesthetic work, and industrial design directed at appearance rather than a technical characteristic.
• Scaling an existing process by buying more of it.

The general limits are in what does not qualify.

Working to a customer’s specification

This is the question that decides most engineering claims, and it has two answers depending on the period.

For accounting periods beginning on or after 1 April 2024, the merged scheme asks who intended and contemplated the R&D — broadly, the party that decided the work was needed and took the risk of it. Where a customer specifies an outcome and leaves the method to you, the claim is usually yours. Where a customer commissions the development itself, it is usually theirs. The full test, which is more involved than that summary, is in contracted-out R&D, and it remains the most contested area in the relief.

For earlier periods, which still matter on open years, amendments and enquiries, two First-tier Tribunal decisions in late 2024 went the taxpayer’s way on facts that describe a great many engineering businesses.

In one, a specialist fit-out contractor had agreed to deliver defined works for a fixed price and found during delivery that standard solutions would not do — acoustic encasements, an unusual brick detail — so it developed its own. The tribunal held the expenditure was not subsidised, because the bargain was to deliver the works for a price and not to reimburse particular costs, and not contracted out, because the company kept the resulting knowledge and bore the economic risk. It noted the purpose of the contracted-out rule is to stop the same R&D being relieved twice by passing the claim up the chain, not to deny relief to a contractor solving its own problems.

In the other, an engineering and automation business taking commissions for one-off structures succeeded on the same two points. The reasoning is worth knowing because it is so recognisable: clients bought a specified end product at a fixed price, often did not know whether R&D would be needed, and left the method entirely to the supplier. The tribunal adopted an analogy that puts it better than any statutory paraphrase — the customer wanted the picture on the jigsaw box, and had no interest in how the pieces went together.

The same business drew the line honestly in the other direction: where a client asked it to develop a specific technology, it made no claim, treating that as sub-contracted R&D. That distinction — a specified deliverable versus a commissioned development — is the one to apply to your own contracts, whichever period you are in. What it does not turn on is what the contract is called.

Prototypes, first articles and pilot plants

R&D ends when the uncertainty is resolved. In this sector the practical markers are reasonably clear, and they rarely coincide with a project’s commercial milestones.

Designing, building and testing a prototype is generally R&D, but once modifications reflecting the test findings are made and further testing is satisfactorily completed, the uncertainty is resolved, and later work is not. Building and running a pilot plant is R&D “until the scientific or technological uncertainty associated with the intended advance in science or technology has been resolved” — the plant may keep running long after the claim has ended.

First articles, pre-production runs and trial batches sit on that line. Where the trial is resolving whether the process works, it is inside; where it is demonstrating to a customer that a settled process meets their requirement, it is outside. And a new problem emerging later in production can start a fresh project, provided it is genuinely a new uncertainty rather than fault fixing.

Costs that behave differently in this sector

The general position is in what costs qualify. Two points catch engineering businesses in particular.

Material in something you sell is excluded. Where the R&D produces an item that is then transferred for consideration in the ordinary course of business, the cost of the consumable items forming part of it is not qualifying expenditure. The prototype that becomes the delivered article, and the first-article parts invoiced to the customer, are the usual casualties. Where only part of a batch is sold, an apportionment applies, so material scrapped or retained for further trials stays in.

The accounts do not decide capital and revenue. Tooling, jigs, fixtures, and rigs built for a development project need to be assessed by their nature rather than where the bookkeeping puts them, and capital spend that falls outside the claim may have a home in research and development allowances instead. Capitalising development costs in the accounts does not by itself make it capital for tax.