Is Your Project R&D? What Qualifies for Tax Relief

When people hear “R&D”, they picture lab coats and clinical trials. That’s understandable, but it’s a long way from what HMRC means.

If you’ve ever solved a hard technical problem in your business – built software that pushed a system somewhere new, worked out a manufacturing process nobody had cracked before, found a way to build something with materials that hadn’t been used that way – there’s a fair chance you’re doing R&D in the tax sense without ever calling it that.

The trouble is, the rules tightened in 2024, and HMRC is no longer waving claims through as it used to. Claim volumes have fallen by around a quarter, and weakly evidenced claims are getting picked apart in ways they weren’t a few years ago. So knowing where the line falls between qualifying work and routine development matters more than it used to.

This guide walks through what counts as R&D for tax in 2026, the three questions HMRC asks, where claims tend to come unstuck, and how the two schemes work now.

What HMRC’s Looking For

HMRC follows the DSIT guidelines – they were called the BEIS guidelines until the department was renamed in 2023. Strip them back and you’re left with three questions, and your project has to be a yes on all three.

Did the work try to advance science or technology?

Not just within your company, but within the field as a whole. If someone with the right expertise could have looked up the answer or worked it out using methods already documented and shared, then it’s not R&D. The advance has to be in what’s publicly known.

Was there meaningful technical uncertainty?

The kind a competent professional couldn’t readily resolve going in. Could they have predicted whether the outcome was achievable, or known how to achieve it? If yes, you’re not in R&D territory. Commercial uncertainty (will customers buy it?), financial uncertainty (can we fund it?) and deadline pressure (can we ship by Friday?) don’t count, however stressful they feel.

Was the work systematic?

Methodical, recorded, properly structured. Iteration counts. Prototyping counts. Even unsuccessful attempts count, where they’re part of a coherent effort to resolve the uncertainty. What doesn’t count is tinkering without a plan, routine testing that doesn’t address an uncertainty, or just tweaking how something looks.

That middle test is where most edge cases turn. Technical uncertainty has a specific meaning: it’s about whether something can be done at all, or how. Not about whether it’s easy.

What Counts as R&D in Practice

R&D doesn’t belong to one sector. Here are some areas where claims regularly hold up, provided the work and documentation match the test above.

Software Development

Software is one of the biggest sources of R&D claims. It’s also one of the areas where qualification is most often missed, because developers tend to think of their work as engineering rather than research.

But “engineering” and “R&D” aren’t mutually exclusive. If you’re solving technical problems that don’t have a documented answer – pushing performance beyond what existing tools handle, integrating systems in ways nobody’s tried before, building algorithms where existing ones don’t fit – you’re doing R&D in the tax sense.

Common qualifying activities include:

• Performance work: Building software to handle workloads, latency or data volumes that current tools don’t cope with comfortably.
• Novel integrations: Combining systems where the combination isn’t documented and the interactions create unknowns.
• Algorithm development: Designing new algorithms, or adapting them in ways that aren’t already mapped out.
• Platform challenges: Making something work reliably in an environment where the technical compatibility wasn’t obvious going in.

What doesn’t qualify is routine work – building a standard website, configuring an off-the-shelf CRM, applying known UI patterns, or translating business requirements into code.

Manufacturing and Process Innovation

Process R&D often makes cleaner claims than product R&D. The uncertainty is concrete – you can usually point at a specific technical question you couldn’t answer at the start – and the documentation tends to exist already, because engineering teams keep test logs and change records as a matter of course.

Things that regularly qualify:

• New manufacturing methods that improve yield, cut waste, or work at temperatures, pressures or speeds existing equipment can’t comfortably handle
• Materials development – new alloys, composites, polymers, or familiar materials in unfamiliar contexts
• Quality and tolerance challenges where the route to a tighter spec isn’t obvious
• Scale-up problems taking a lab or pilot process into full production and running into engineering hurdles you couldn’t have predicted

Construction and Engineering

Construction used to be a quiet area for R&D claims. It’s grown sharply over the past few years as the work has been better understood.

What tends to qualify:

• Novel structural methods where standard engineering practice doesn’t resolve the uncertainty
• Material substitutions at scale – sustainable, recycled or unfamiliar materials in load-bearing or critical applications
• Site-specific engineering where ground conditions, environmental constraints or planning restrictions demand fresh technical work
• Building services innovation – HVAC, electrical, fire safety systems pushing beyond standard solutions

What doesn’t qualify is the routine application of established methods, no matter how big or expensive the project.

Food, Pharmaceuticals and Biotech

This is the area closest to what people visualise when they hear “R&D” – the lab coats, the test protocols. The reality stretches further than just drug development.

Food businesses often qualify. Reformulating a product to remove an allergen, hit a clean-label position, or extend shelf life can throw up material technical questions about whether the new version will hold up. If the route to the result isn’t obvious to a competent food technologist, the work fits the test.

Pharma and biotech R&D was the original use case for the relief, and remains the most well-established. Credible claims in those sectors are usually accepted, provided the underlying paper trail holds up.

Professional Services

This is the toughest sector to qualify in.

Tax advice, legal work, marketing strategy, financial planning – almost none of it is R&D in the statutory sense, however clever or original it feels from the inside. The work is intellectual, but it doesn’t usually involve the kind of scientific or technological uncertainty the relief is aimed at.

The exception may be where a professional services company builds technology – their own software platform, a new testing methodology, an analytics engine. If that work meets the three tests on its own merits, it can qualify.

Where Claims Tend to Come Unstuck

HMRC has turned down more claims in recent years, and the same patterns recur:

• Commercial novelty mistaken for technological uncertainty: A new product isn’t R&D just because it’s new to market. The question is whether the technology behind it had to push beyond what was already known.
• Routine engineering dressed up as R&D: Long, hard projects feel like R&D from the inside. But if a competent professional could have seen the route from problem to solution, it doesn’t qualify.
• Thin documentation: Claims need records that were made as the work was happening – design notes, test results, technical reports, version control logs. A narrative written up after the fact won’t hold up. The Additional Information Form needs a senior internal contact and the project descriptions get scrutinised line by line.

The cost of getting it wrong has gone up too. Where a claim falls apart, HMRC now routinely claws back any credits paid out, with interest and sometimes penalties on top.

The Schemes in 2026

The R&D landscape changed in April 2024. Two schemes replaced the old SME and RDEC reliefs, and most companies now claim under one or the other.

Merged Scheme RDEC

For accounting periods starting on or after 1 April 2024, this is the default scheme for most companies.

How it works:

• A 20% gross credit on qualifying R&D spending, treated as taxable income for the company.
• What you keep: After corporation tax on the credit itself, around 15% of qualifying spend at the 25% main rate, or 16.2% at the 19% small profits rate.
• Loss-making companies can take the credit as a cash refund instead. The cash element is capped at £20,000 a year plus three times the company’s annual PAYE and NIC bill. Anything above the cap carries forward to next year’s claim. The £20,000 floor is scaled down for accounting periods shorter than 12 months.
• Open to companies of all sizes, with one exception below.

ERIS for Loss-Making R&D-Intensive SMEs

Enhanced R&D Intensive Support is a more generous regime for SMEs whose business is primarily R&D.

To qualify, you need to:

• Meet the SME definition – fewer than 500 staff and either turnover under €100m or balance sheet under €86m, counting any linked and partner companies
• Be loss-making for tax purposes in the period
• Spend at least 30% of total expenditure on qualifying R&D (down from 40% before April 2024)

ERIS lets you deduct 186% of qualifying spend from profits instead of the usual 100%. Where that creates a loss, you can convert it into a cash payment from HMRC at 14.5p in the pound. Combined, that’s worth roughly 27% of qualifying spend – substantial money for an R&D-intensive startup.

You choose between the merged scheme and ERIS each accounting period, so a company can move between them as its profile changes.

ERIS also counts towards wider limits on the total subsidy a company can receive over a rolling three-year window, and Northern Irish companies face additional rules – both of which are worth checking before submitting a claim.

The Claim Process

The procedural side of R&D claims has tightened. The table below sets out the sequence for any claim in 2026.

The Additional Information Form is the document most claims live or die by. It covers the company details, a write-up of each project (including the technical advance, the uncertainty, and the work done to resolve it), the breakdown of qualifying spend by cost category, and contact details for the senior person within the company responsible for R&D, plus any agents involved.

Our R&D tax credits article covers more on who qualifies and how to assess whether your business has a claim worth making.

How Double Point Can Help

R&D tax relief still rewards meaningful innovation with serious cash. But it punishes overreach harder than it used to – rejected claims, repaid credits, interest, sometimes penalties.

This guide is general information only, not personal tax advice. Whether your specific project qualifies depends on the technical detail, the documentation, and the wider position of the company.

At Double Point, our chartered accountants help limited companies across the UK get R&D claims right – running technical reviews to identify qualifying work, preparing the Additional Information Form, modelling the choice between the merged scheme and ERIS, and supporting any HMRC enquiry that follows.

Book a free consultation and we’ll look at whether your work qualifies for R&D relief in 2026 and the cleanest way to claim it.