Your company might already be doing R&D. Here's how to tell

Ask the founder of a typical 40-person engineering firm whether the company does research and development, and the answer is usually a polite no. There's no lab, no R&D department and no patents. The team "just solves problems for customers."

That answer is sometimes right. But it often comes from a picture of R&D that bears little relation to the definition used for UK tax relief. The definition doesn't care about job titles, white coats or whether the work was a success. It asks a narrower and more interesting question: was the company trying to do something that a competent professional in the field couldn't readily work out how to do?

That question is worth asking properly. The UK's R&D reliefs exist to reward exactly that kind of work. Yet some established companies never check, because they assume the word "research" doesn't apply to them.

What the definition actually says

The government's guidelines on the meaning of R&D for tax purposes come down to three things.

First, the project has to seek an advance in science or technology. As the guidelines put it, the advance has to be in the overall field, not just for one company. Building something that's new to you but routine for the industry doesn't count. Neither does using technology to make a product: almost every product uses technology.

Second, there has to be scientific or technological uncertainty. In other words, a competent professional working in the field couldn't readily work out whether the thing was possible, or how to achieve it, from knowledge that was already available.

Third, the company has to do work to resolve that uncertainty: investigating, testing, building and trying again.

Two details surprise people. The project doesn't have to succeed; work that seeks an advance can still count even when the advance isn't achieved. And uncertainty can sit in the late stages, when something already shown to work in principle has to become reliable, repeatable and affordable at scale.

Three everyday examples

A software company with a performance problem. A business software firm has a client whose data volumes have outgrown the system. Standard approaches such as adding servers, tuning the database or caching results don't get processing times anywhere near what's needed. The developers spend months testing different ways of structuring and processing the data, and most of them fail.

Building the client's new screens and reports wouldn't be R&D. That's skilled but familiar work. The months spent working out how to hit a performance level that existing methods couldn't reach may be worth a closer look.

A food manufacturer changing a recipe. A producer needs to cut the sugar in a product line, but sugar does more than sweeten. It affects texture, browning and shelf life. Swapping in an alternative changes how the product behaves on the line and on the shelf, and the team runs trial after trial to get a version that holds up for the full shelf life.

Redesigning the packaging wouldn't count, because a change to look or customer appeal isn't a scientific advance. Resolving how to keep the product stable without the ingredient that used to do that job may be.

An engineering firm pushing a component further. A precision engineering company is asked for a part that has to perform at temperatures and loads beyond what its usual materials and methods handle. Nobody on the team, or in the published material they can find, can say whether it's achievable with the processes available. The firm tests materials, finishes and designs, and scraps several prototypes.

Buying a newer machine wouldn't qualify on its own, and neither would routine quality testing. The investigation into whether and how the part could be made to perform may.

None of these companies would describe themselves as research businesses. All three may have done something that meets the definition.

What usually doesn't count

The boundaries matter as much as the examples:

• Routine work: copying or adapting an existing product or process, even if it's new to your company.
• Existing technology in a new setting: applying existing technology to a new trade with only minor or routine changes.
• Fine-tuning: optimising or tweaking that doesn't touch the underlying science or technology.
• Commercial work: marketing, production, distribution and general administration around a new product.
• The wrong fields: work in the arts, humanities and social sciences, including economics.

The records that help

If you think a project might qualify, the most useful thing you can do is keep a simple record while the work happens, not reconstruct it a year later. In practice that means:

• The problem at the start. What were you trying to achieve, and why couldn't the team, or the published knowledge in the field, simply tell you how?
• What you tried. The approaches tested, including the ones that failed. HMRC's guidance says you need to be able to explain the work done, including what succeeded and what didn't. Notes made at the time make that much easier.
• Who did the work, and for how long. Staff time is often the biggest cost in a project, and a reasonable split of time spent on the uncertain work is far easier to support when it's recorded as you go.
• What it cost. Materials and consumables used up in testing, relevant software, and some subcontracted work (the rules on contractors are specific, so check them), kept separate from routine spending.
• The dates. When the uncertainty was first tackled, and when it was resolved or the work stopped.

The process has also become more formal. A company that is claiming for the first time, or hasn't claimed in the past three years, may need to tell HMRC in advance that it plans to claim. Claims also need an additional information form describing the projects, submitted before the claim itself. Both are tied to the company's accounting period, so check the dates early rather than discover a missed window later.

Ask the question properly

None of this means every growing company is sitting on an R&D claim. Many aren't, and a good adviser will say so. Whether a project meets the definition depends on its facts, and HMRC decides whether a claim is accepted.

But "we don't do R&D" deserves the same scrutiny a founder would give any other assumption about the business. The question isn't whether you have a lab. It's whether your team spent time on a technical problem it genuinely didn't know how to solve. If the honest answer is yes, it's worth finding out whether that work counts.

Author bio

Mike Simmons is a Director at The Business Review Co., which helps established UK businesses recover and reduce expenditure. Its guide to what counts as R&D for tax relief summarises the government's definition.

‍