What evidence does HMRC want for an R&D claim?

HMRC's starting point is the additional information form, now mandatory for claims covering accounting periods beginning on or after 1 August 2023, which requires a breakdown of qualifying costs and a description of the projects, including how many were included and the technical uncertainties addressed.

Beyond the form, HMRC wants detail showing genuine engineering, scientific or software development uncertainty — not simply that something was difficult, expensive, or new to the company. Evidence such as design iterations, failed attempts, technical specifications, and internal discussions about how to solve a problem carries real weight.

On the cost side, HMRC checks that staff time claimed as qualifying is properly apportioned using a reasonable methodology, that subcontractor and externally provided worker costs meet the connected or unconnected party rules, and that consumables genuinely relate to the R&D activity.

Under the general principles applied by HMRC and tribunals in R&D cases, evidence created at the time the work was carried out is far more persuasive than a narrative reconstructed after the event purely to support a claim. Contemporaneous records demonstrate that the uncertainty genuinely existed and was worked through, rather than being described with hindsight.

This is reinforced by how HMRC assesses claims under Part 13 Corporation Tax Act 2009 (opens in a new tab): the legal test focuses on what a competent professional could not have readily deduced at the time, so evidence must speak to that point in time.

Where a company cannot produce contemporaneous technical records, it can still build a case using witness evidence from the staff involved, but this is generally weaker and more open to challenge than documents created as the work happened.

Put a simple system in place going forward: even brief notes made at the time a technical challenge is identified and worked through are far more valuable later than a detailed narrative written months afterwards.

Keep timesheets or a defensible time-apportionment method for staff working partly on qualifying projects, since HMRC frequently challenges cost apportionment even where the technical case is accepted.

If you are already under enquiry and contemporaneous records are thin, focus on the strongest and most defensible projects rather than trying to justify every line of a broad claim equally.

Building an evidence base from scratch during an enquiry, rather than having kept it as you went, takes considerably more adviser and staff time, and the outcome is less certain.

Growth plan clients have free tax investigation insurance included, covering our fees for helping assemble and present evidence during an R&D compliance check — see /fees.

Investing modest time in record-keeping at the point work is done is far cheaper than reconstructing a defence years later.

Do not rely on a single company-wide narrative for multiple distinct projects; HMRC expects project-level detail.

Do not submit a claim without the mandatory additional information form where required — claims without it can be removed from the return entirely.

Do not assume enthusiastic marketing language about innovation is the same as evidence of qualifying technical uncertainty; HMRC looks past sales copy to the underlying facts.

Set up a simple running log per project capturing the technical challenge, the approaches tried, and the outcome, updated as the work happens rather than at year end.

Record staff time against projects using a consistent, explainable method, even if it is not minute-by-minute timesheeting.

Review your evidence file before each claim is submitted, not just before an enquiry starts, so gaps are caught early.