R&D tax relief •10 min read

Realbuzz Group Ltd v HMRC

MJ

Matthew Jones ACA CTA

Last reviewed September 2026 · Editorial standards

Tribunal
First-tier Tribunal (Tax Chamber)
Decided
1 May 2025
Judge
Judge Marilyn McKeever and Mr Mohammed Farooq

The company won, and the merits had nothing to do with it. Realbuzz accepted that no R&D relief was due on a claim of £335,452.57 and that more tax should have been paid. It kept the money anyway, because HMRC had let the enquiry window close on a claim a competent officer should have realised was excessive. It is the only decision we have found on how HMRC reopens an R&D year, and it turns on procedure throughout.

What was at issue

Realbuzz Group Ltd was established in 2000 and works in the fitness, sports and running sectors. Its activities are online content and social networking, and — the commercially important part — online entry systems for mass-participation events. Those events include the London Marathon, the charity fundraising programmes attached to them, and “virtual events” completed in the entrant’s own area (paragraph 18). It was an SME, and the decision records the applicable rate of relief for the year as 230% (paragraph 19).

The claim was for £335,452.57 for the accounting period ended 30 April 2020, made in an amended return received on 31 March 2021 and supported by an R&D report the company’s accountants prepared after discussions with its technical team (paragraphs 1 and 6). HMRC raised a discovery assessment on 1 June 2023 disallowing it in full.

By the hearing, only procedure was in dispute. At paragraph 55 the tribunal recorded: “It is common ground that no R&D relief was due for APE 2020 and that additional tax should be due.” HMRC also accepted the return had not been completed carelessly, which closed the alternative route into the year (paragraph 56). Everything rested on paragraph 44(1) of Schedule 18 to the Finance Act 1998, and on a secondary question: whether the R&D report filed for the following year counted as information made available for the earlier one.

How the dispute got to a hearing

The case is about what HMRC knew and when.

Realbuzz filed its 2020 return on 22 December 2020 and amended it twice, the second amendment on 31 March 2021 adding the R&D claim and the 2020 report. It filed its 2021 return on 13 July 2021, and HMRC opened an enquiry into that later year on 17 September 2021. On 15 November 2021 the accountants sent HMRC the 2021 R&D report, which covered several projects begun in 2019 that had also appeared in the 2020 report (paragraphs 6 and 7).

On 13 April 2022 the officer handling the enquiry wrote saying most of the projects did not fall within the definition of R&D, and listing which. He invited the accountants to explain why they disagreed, and enclosed a questionnaire he intended to pass to HMRC’s software specialists (paragraphs 8 and 103). Seventeen days later, on 30 April 2022, the window for enquiring into the 2020 return closed — the “LDE” (paragraph 9).

A second officer took over in September 2022 with no handover available (paragraph 47). After further correspondence and a conference with HMRC’s software specialists on 20 January 2023, HMRC closed the 2021 enquiry on 8 March 2023 and disallowed that year’s claim in full, unappealed. The same letter said HMRC was considering a discovery assessment for 2020 because the same or similar projects had been claimed (paragraphs 50 and 111).

The assessment was issued on 1 June 2023, appealed on 30 June 2023, upheld on review, and notified to the tribunal in time on 30 November 2023 (paragraphs 12 and 13). It was heard at Taylor House on 28 March 2025 and released on 1 May 2025.

What each side argued

HMRC’s case had two limbs. First, quantification: even if too little tax had clearly been charged, the officer is not barred unless the amount could be worked out. The claim is for a specific sum, and an officer who cannot say how much to assess has not been made aware of “the situation” (paragraphs 61 and 74). Second, complexity: software is a difficult field, the hypothetical officer is not a specialist, and while some projects were obviously non-qualifying, others would have needed further enquiry (paragraphs 88, 89 and 92). From that HMRC argued its primary position — that it was barred only if the officer should have realised no relief at all was due — and an alternative, that at most the obviously bad parts were protected (paragraphs 140 and 141).

For the company, it was argued that quantification is no part of the test: the hypothetical officer must be aware of an actual insufficiency, and the amount is the real officer’s job (paragraph 80). On complexity the company pointed to the authorities treating it as a feature of difficult law rather than difficult facts (paragraph 94). Its primary submission was that if the officer should have been aware the relief was excessive, HMRC is shut out of the whole claim; its alternative, that at least the adequately disclosed parts are protected (paragraphs 143 and 144). Allowing inadequate disclosure on some parts to open up the whole would strip the provision of the protection it exists to give (paragraph 142).

The provision the case turns on

Paragraph 44(1) of Schedule 18 to the Finance Act 1998 fixes the moment at which the question is asked: the time when an officer “ceased to be entitled to give a notice of enquiry into the return”. A discovery assessment may be made only if, at that time, the officer “could not have been reasonably expected, on the basis of the information made available to him before that time, to be aware of the situation mentioned in paragraph 41(1) or (2)”. Paragraph 41(1)(c) covers the case where “relief has been given which is or has become excessive”.

Paragraph 44(2) then defines information made available, in four limbs. The first is information “contained in a relevant return by the company or in documents accompanying any such return”. The second is information “contained in a relevant claim made by the company or in any accounts, statements or documents accompanying any such claim” — the limb an R&D claim turns on. The third is information produced for the purposes of an enquiry. The fourth is information whose existence and relevance “could reasonably be expected to be inferred” from those sources, or is “notified in writing” to an officer. Legislation.gov.uk records no outstanding effects for paragraph 44.

The burden sat on HMRC to prove the assessment valid (paragraph 63). The tribunal took its principles from HMRC v Fisher [2021] EWCA Civ 1438, drawing on Veltema, Lansdowne, Sanderson and Beagles. The officer is hypothetical, of general competence and with a reasonable knowledge of the law. That officer must be clearly alerted to an insufficiency rather than merely prompted to enquire, and may use only the specified information (paragraphs 66 and 67).

What the tribunal decided

The appeal was allowed (paragraph 162), on three findings.

On quantification, the tribunal preferred the company: “the hypothetical officer does not have to quantify the amount of the tax underpaid/relief overclaimed in order to be ‘aware of the actual insufficiency’ in the assessment. He has only to be aware that there is an actual insufficiency” (paragraph 86). Fixing the amount is the real officer’s task, on all the information gathered later.

On complexity, the tribunal accepted that the hypothetical officer is no software specialist but held that “the law relating to R&D applicable in this case is relatively straightforward” (paragraph 96), and that what matters is the adequacy of the company’s own disclosure. Reading the 2020 report, three projects were plainly outside the relief and “no special technical knowledge was required to come to that conclusion” (paragraph 129). It corrected HMRC on one point along the way: “a more efficient product or cost improvements can be R&D within paragraph 7 of the Guidelines … Making something faster and more efficient can be an appreciable improvement” (paragraph 121).

On the decisive question, the tribunal found that the officer need not have seen that the whole claim failed. “The hypothetical officer would not have been expected at the LDE to be aware that the entire claim was non-qualifying, but he would have been expected to be aware that some of the projects and sub-projects did not qualify for relief” (paragraph 133). Because a single claim for a single sum is a single inaccuracy, that was enough. The finding was framed as a rule: “Where there is a single claim to relief which is excessive because some parts of it qualify and other parts do not” (paragraph 147). Where the hypothetical officer at the LDE “should reasonably have been aware that the claim for relief was excessive, we consider that paragraph 44(1) prevents HMRC from raising a discovery assessment in relation to that claim” (paragraph 147). HMRC’s reliance on Hargreaves and Lynch failed: those cases concerned two separate charges, not one (paragraph 146).

The secondary point went against the company, though it made no difference. The 2021 report “was not information made available to the hypothetical officer” (paragraph 160). The 2020 report’s statement that projects were “ongoing” would not lead an officer to infer that a later report existed. Awareness of a problem in 2021 would raise only a suspicion worth investigating, which the hypothetical officer may not act on (paragraphs 155 and 158).

One point for anyone quoting the decision: paragraph 112 reproduces HMRC’s stated basis for the assessment, referring to accounting periods ending “31 April 2020” and “31 April 2021”. April has thirty days.

What it changes for a claim being prepared now

  • A closed year is not automatically closed. Discovery is the route HMRC uses once the enquiry window has passed, and paragraph 44 is the only thing in its way. Where an enquiry into a later year touches projects that also ran in an earlier one, expect the earlier year to be looked at.
  • Disclosure is the defence, and it is built at filing. The protection turns on what a competent officer could have concluded from the return, the claim and the accompanying documents. A report that states each project’s aim and outcome plainly enough for a non-specialist to judge does two jobs: it supports the claim, and it closes the year.
  • Date the enquiry window and diarise it. Seventeen days decided this case. Record, for every claim, when the return was delivered and when HMRC’s right to enquire expires. Can HMRC make me pay back an R&D tax credit? sets out the routes back into a paid claim.
  • Conceding the merits does not concede the assessment. The company accepted the relief was not due and still succeeded. Whether an assessment is barred is a separate argument, and it survives a concession on the technology.

This is an old-scheme decision on the relief — a 230% additional deduction for a period ended 30 April 2020 — and the current schemes replace that arithmetic for accounting periods beginning on or after 1 April 2024. The procedural rules are untouched: paragraph 44 applies to a merged-scheme or Enhanced R&D Intensive Support claim exactly as it applied here. What has changed around it is the paperwork. An additional information form is now required for every claim, and advance notification for a company that has not claimed in the three years ending with the notification deadline. There is more on HMRC’s file at the outset, and correspondingly more that a hypothetical officer could be expected to have read.

Where it sits against the other decisions

It pairs with Stage One Creative Services, the other decision in which an R&D claim was defended on how the year was opened rather than on whether the work qualified. Together they make the same point from two directions: the technical merits are not the only thing an enquiry has to get right. It is also the counterpart to Assembly Global Networks Ltd v HMRC [2024] UKFTT 304 (TC), where an enquiry letter drafted sixteen days before the window closed was held validly served. Realbuzz is what happens when the window is missed instead.

Has it been appealed?

No onward appeal has been reported, and no Upper Tribunal decision naming Realbuzz Group appears on the published record. Applications for permission to appeal are not consistently published, so the absence of a reported appeal is not proof that none was sought.

HMRC has not changed its published guidance as a result: no entry attributed to this decision appears in the CIRD manual’s dated change log, and it is not discussed in the published Research and Development Communication Forum minutes.

The decision is First-tier Tribunal, so it binds only the parties and sets no precedent. Its value is that it is the only treatment of paragraph 44 in an R&D context we have found, and it reads across to every claim where HMRC arrives after the window has shut. The rest sit in the register of R&D case law.

Sources

First-tier Tribunal decisions bind only the parties to them and set no precedent. This entry describes the decision as it stood at the review date above, as general information rather than advice on your circumstances — see our terms. For an answer on your own facts, talk to us.