R&D Tax Credits
for startups

Specialist R&D tax credit claims for innovative UK businesses. Expert identification of qualifying activities and maximum credit optimisation for software development, tech innovation, and research projects.

Under the merged R&D scheme effective for accounting periods beginning on or after 1 April 2024, all UK companies claim a 20% above-the-line credit on qualifying R&D expenditure, with R&D-intensive SMEs (where R&D is over 30% of total spending) claiming an enhanced 27% rate. Loss-making startups can elect to surrender the credit for a cash payment from HMRC, often arriving 4-6 weeks after a well-prepared claim.

Claim quality matters more than ever. HMRC's R&D enquiry rate has risen sharply since 2022, with around 1 in 5 SME claims now subject to enquiry. The new mandatory Additional Information Form, pre-notification requirement for first-time claimants, and tightened evidence standards mean DIY claims and generalist-prepared claims are now genuinely risky. Specialist preparation matters.

The April 2024 reform of the UK R&D tax relief regime is the most significant change to startup tax in a decade, and it changed the calculus for almost every claimant. The previous SME scheme provided an enhanced deduction of up to 230% on qualifying R&D expenditure, with loss-making companies able to surrender enhanced losses for a cash payment at the SME scheme rate of up to 14.5%. The new merged scheme provides a single 20% above-the-line credit applicable to all companies regardless of size, with R&D-intensive SMEs (R&D spending over 30% of total spending, threshold reduced to 30% from 40% from April 2024) qualifying for an enhanced 27% rate. The headline rate is lower for many smaller claimants under the new scheme, but the simplified mechanics, cleaner interaction with grant funding, and clearer treatment of subcontractor expenses have made claim preparation more straightforward.

Qualifying R&D activity is defined by HMRC's definition (technically the BIS definition, adopted by HMRC) as work that seeks to achieve an advance in science or technology by resolving scientific or technological uncertainty that wasn't readily resolvable by a competent professional in the field. The key tests are advance (something new or improved beyond the existing baseline of public knowledge), uncertainty (the outcome wasn't predictable from existing knowledge), and competent professional (someone qualified in the field couldn't have known the answer in advance). For software companies, qualifying activity typically includes designing novel algorithms, solving non-trivial integration problems, building new architectures for scale or performance, and producing measurable improvements in functionality or efficiency that go beyond routine application of existing techniques. Activities that don't qualify include routine debugging, cosmetic UI changes, integrating off-the-shelf libraries via documented APIs, and building software using established techniques even if the specific application is new.

Qualifying expenditure under the merged scheme covers staff costs (the proportion of salary, employer NIC, and pension attributable to R&D activity, supported by timesheets or other contemporaneous evidence), externally provided workers (with restrictions on overseas costs introduced from April 2024), subcontractor costs (where the subcontractor performs R&D for the claiming company), software directly used in R&D, consumables transformed in the R&D process, and certain payments to qualifying bodies for R&D services. The post-April-2024 restriction on overseas costs is a material change: where R&D is performed by subcontractors or EPWs based outside the UK, the costs are generally restricted unless specific overseas work conditions apply. UK companies with India-based development teams, for example, have seen claim values reduced by 30-50% under the new rules.

The Additional Information Form, mandatory for all claims with accounting periods ending after 8 August 2023, requires claimants to provide structured information including the company's qualifying activity description, project-by-project breakdown of qualifying expenditure, identification of competent professional, and detail on how each project meets the definition of R&D. The form is the single most important defensive document for an R&D claim - HMRC's enquiry team uses it as the entry point for any review, and gaps or inconsistencies in the form are typical triggers for opening a check. Pre-notification, mandatory for first-time claimants and for any company that hasn't claimed in the previous three years, must be submitted to HMRC within six months of the end of the accounting period being claimed. Missing the pre-notification deadline disqualifies the claim entirely - there's no extension or appeal mechanism.

HMRC's enquiry rate on R&D claims has changed the economics of claim preparation. Pre-2022, R&D claims were a relatively low-friction process with most claims processed without enquiry. Post-2022, HMRC has materially increased its compliance team and enquiry rate, with around 1 in 5 SME claims subject to formal enquiry and a noticeable shift in HMRC posture from sympathetic to sceptical. Common enquiry triggers include claims that look anomalously large for the company size, claims in sectors HMRC perceives as low-genuine-R&D (digital marketing, ecommerce platform setup, basic web development), claims with thin technical narratives or generic descriptions, claims by first-time claimants without specialist advisors, and claims where the cost methodology appears unsupported. A good R&D advisor maintains a defensible position on each of these points, with contemporaneous timesheets, project documentation, and a clear technical narrative ready for the file.

Cash flow timing on R&D claims is the operational point that matters most for pre-revenue startups. A loss-making SME claiming under the merged scheme can elect to surrender the credit for a cash payment from HMRC. HMRC's published service standard is to process credit payments within 28 working days of receipt of a complete claim, but in practice the timeline varies from four weeks for clean claims to six months or more where an enquiry is opened. For pre-revenue companies depending on the R&D credit as the largest single annual cash inflow, the timing of submission, the quality of preparation (which directly affects enquiry probability), and the company's working capital headroom in case of a delayed payment all matter. Submitting the claim in the first month after year-end maximises the cash benefit; waiting six months adds material risk. Specialist R&D advisors typically build the claim alongside the year-end accounts so submission is possible within four weeks of accounting period end.

Software R&D claims have been the area of most active HMRC scrutiny since 2022. The qualifying-activity test for software is structurally harder to demonstrate than for hardware or wet-lab science because software work always uses existing languages, frameworks, and libraries, making it easier for HMRC to challenge whether the work represented a genuine advance beyond established techniques. Successful software R&D claims typically demonstrate technological uncertainty in one of: novel algorithms (genuinely new mathematical or statistical methods), performance or scale beyond what existing techniques can deliver (10x or more, with measured baselines), architectural advances (genuinely new patterns for handling concurrency, distribution, or fault tolerance), or AI/ML model design where the model itself represents an advance. Claims framed around 'building a CRUD app with React and Postgres' or 'integrating Stripe and Salesforce' will not survive a competent enquiry, regardless of how much development work was actually done.

Subsidised R&D under the new merged scheme has different rules than the old SME scheme. Under the old rules, costs subsidised by notified state aid grants (Innovate UK, etc.) were excluded from SME relief but could be claimed under the older RDEC scheme at lower rates. Under the merged scheme, this distinction has been simplified: the merged credit treats most arrangements neutrally and the company can typically claim on the full project costs including the grant-subsidised portion. The exception is for certain notified state aid arrangements that explicitly conflict with the merged scheme, which mostly applies to specific bilateral grant programmes rather than mainstream Innovate UK funding. The accountant should review each grant agreement to identify any such exclusions, but for most companies the practical answer is that grants and R&D credits stack cleanly under the merged scheme.

Group company R&D claims and intercompany work introduce specific complexities. Where R&D is performed by one group company on behalf of another (a holding company with a subsidiary trading entity, for example), the claim is typically made by the company conducting the R&D, not the company funding it. Where R&D is contracted to an unconnected subcontractor, the claim sits with the company commissioning the work, with restrictions on the proportion claimable. Where R&D is performed by an overseas group entity, the post-April-2024 overseas-cost rules apply, generally restricting the claim. Internal transfer pricing arrangements between group companies have to be reviewed for any R&D claim, because misaligned transfer pricing can move the qualifying expenditure to the wrong entity and create both R&D denial and transfer pricing risk.

Pivot scenarios and changing project descriptions over a claim period need careful documentation. A startup that pivoted mid-year from product A to product B has two separate qualifying activity narratives, with the work on product A counting up to the pivot date and the work on product B counting from the pivot date. The competent-professional and uncertainty tests apply separately to each. A common error is to retroactively reframe the abandoned project as a successful learning exercise to bolster the claim, which HMRC sees through quickly because the contemporaneous documentation (commits, design docs, board updates) doesn't match the post-hoc narrative. Honest contemporaneous documentation is the strongest defence; reconstructed narratives are a weakness.