How the Winemakers Guide Corked an R&D Claim

Two First-tier Tribunal decisions, September 2026 · 1 Oct 2026

Two companies lost their R&D tax relief appeals at the First-tier Tribunal in September 2026, three weeks apart, and in neither case was the company accused of making anything up. Both had done real technical work. Both could explain, in detail, the problem they had been trying to solve. Both lost because they could not establish what the existing knowledge in their field could and could not already do.

In the first of them, the Tribunal settled the point by looking something up itself. The company had been developing canned wine packaging. Nobody had put the Winemaker’s Guide in evidence; the Tribunal consulted it anyway, found that the technique at the centre of the claim was already described there, and that was effectively the end of the advance argument.

That is a useful thing to know before you file, and an uncomfortable thing to discover afterwards. What follows is what each decision actually turned on, the single distinction that runs through both, and the one costs point in them that can change a figure rather than just weaken an argument.

RDM Wines: three findings, each fatal on its own

RDM Wines Limited v HMRC [2026] UKFTT 1348 (TC) was released on 24 September 2026. The company had claimed relief for work on canned wine packaging, carried out with a South African wine producer. The appeal was dismissed. Three strands of the reasoning matter well beyond the drinks trade.

Novelty is not the test. At paragraph 87 the Tribunal held that “uniqueness, or novelty, is not the relevant test of whether something amounts to R&D.” This is worth dwelling on, because being first is how most people instinctively describe innovation, and it is how a great many claim narratives are written. The statutory question is whether an advance in overall knowledge or capability was sought, and whether scientific or technological uncertainty had to be resolved to get there. A thing can be entirely new to the company, new to its customers, even commercially unprecedented, and still not be an advance in the field.

Ordinary industry literature sets the bar. Having found that the Winemaker’s Guide already “discusses using nitrogen gas to fill a headspace”, the Tribunal concluded at paragraph 86 that activities falling within the practices that the guide describes are not advances. The company needed to show the resolution of uncertainty leading to an advance beyond what a competent professional would readily deduce, and could not. The guide was not expert evidence and was not filed by either side. It was simply the sort of reference material anybody working in that field would be expected to know.

The competent professional is a test, not a job title. At paragraph 32, citing Flame Tree Publishing, the Tribunal adopted the test that a competent professional must be able to demonstrate “appropriate qualifications, experience and up-to-date knowledge of the relevant scientific and technological principles involved.” The third limb did the real work here. A named individual with a long tenure satisfies the first two and tells the reader nothing about whether that person’s knowledge of the field is current — and currency of knowledge is what the advance argument rests on, because it defines what the field already knew.

Documents beat recollection. At paragraph 34 the Tribunal applied Gestmin, on the basis that “memories are fluid and malleable, being constantly rewritten whenever they are retrieved”, and preferred inferences drawn from the contemporaneous documents. It rejected the claimed May 2021 project start, finding on the documents that the relationship with the overseas producer began only in November 2022, and gave little weight to a supporting letter obtained in June 2026, treating it as reconstruction after the event.

It also set out what it would have expected to see and did not:

• written correspondence showing project planning
• records of test results
• internal emails discussing outcomes and alternative approaches
• timesheets or contemporaneous records of staff involvement
• invoices from contractors for the various testing phases
• contract terms for engaging the overseas producer

That list is the most immediately practical paragraph in either decision. None of those items can be created later.

Environmental Services: the same failure, with no handbook involved

Environmental Services Limited v HMRC [2026] UKFTT 1301 (TC) was released two weeks earlier, on 10 September 2026. A waste management company had claimed for two projects — tanker modifications and a storage tank system — in the periods ended 31 July 2020 and 31 July 2021. The appeal was dismissed on both qualification and expenditure grounds.

Operational problem-solving is not R&D. The Tribunal accepted that the company had done “genuine work to improve its business operations.” It still found the work reactive: the “evidence did not establish that ESL’s operational challenges arose because existing scientific or technological knowledge was insufficient.” That sentence is the whole distinction. Difficulty in the business is not the trigger. Insufficiency of knowledge in the field is.

No baseline, no advance. At paragraph 95: “Nor was the Tribunal provided with a clearly articulated technological baseline from which any advance could be measured.” Without that starting point, there is nothing against which an advance can be sized, and the work was characterised instead as adaptations of existing techniques and equipment.

Note what did not happen here. No industry handbook was produced by the parties or by the Tribunal. The claim did not fail because the field turned out to know something the company did not realise it knew. It failed because the field’s state of knowledge was never described at all, which left the panel with a well-evidenced business problem and no way to connect it to the statutory test.

Other panels are picking up this reasoning. At paragraph 70, the Tribunal cites Tanglewood Care Services [2026] UKFTT 1137 on paragraph 9 of the DSIT Guidelines not being read in isolation from paragraph 6, and on advances arising through the resolution of uncertainty affecting capability as well as knowledge. Tanglewood is no longer an isolated decision; it is being applied.

One point on timing. The periods under appeal were pre-reform, but nothing in the reasoning is specific to the old SME scheme or RDEC. The qualification analysis applies the same statutory test and the same DSIT Guidelines that apply now, and the expenditure point rests on CTA 2009 wording that the merged scheme carries forward unchanged.

Both companies could describe their problem. Neither could describe the field.

Strip the two decisions back and the same failure sits underneath both. Each company arrived able to explain, credibly and in technical detail, what it had been struggling with. Neither arrived able to explain what the body of knowledge in its field already contained, and therefore what its work added to it.

The difference shows up in a single sentence of a narrative:

• “We could not get our tankers to handle the new waste stream reliably.” That is a business problem. It may be a hard, expensive, genuinely frustrating business problem, and it is not evidence of anything the Tribunal needs to find.
• “Published knowledge in this field did not establish how to achieve X under these conditions, and here are the standards, handbooks and vendor documentation we tested that against.” That is the start of an R&D claim.

The underlying work can be identical in both versions. Only the second one survives scrutiny, because only the second one engages the statutory test.

This is also why the two decisions are more alarming together than separately. RDM Wines shows what happens when the field’s knowledge is established and the claim turns out to sit inside it. Environmental Services shows what happens when it is never established at all. A claim can therefore fail either by being measured against the literature and falling short, or by offering the Tribunal nothing to measure. The second route is the more common one, and the easier to walk into without noticing.

The costs point: hired equipment is not a consumable

Environmental Services also decided an expenditure question, and this is the one item in either decision that can make a figure wrong rather than make an argument weak.

The company had claimed invoices for equipment hire as consumable items. At paragraph 105 the Tribunal agreed with HMRC that even if the equipment had been used in the R&D, it “was not consumed as part of the R&D process, and as such could not qualify for relief as a consumable.” Consumption is the test. Equipment that is hired, used and handed back is not consumed, and the Tribunal noted the double-relief concern against the separate plant and machinery regimes, which exist precisely to deal with that kind of cost.

It made a broader observation in the same part of the decision: that “numerous invoices appeared on their face to relate to routine maintenance, transport management services, equipment hire, labour, inspections, PPE and other ordinary operational costs.”

That is worth reading as a warning about process rather than about law. Cost categorisation usually happens at the point where a finance team or a bookkeeper sorts a ledger export into claim headings, and “we hired a rig to run the trials” looks, to anyone not thinking about the statutory language, exactly like a cost of doing the R&D. It is a cost of doing the R&D. It is not a consumable item, and the distinction is not intuitive.

If hire, lease or rental lines are reaching a consumables figure in any claim currently in preparation, that is now contrary to a decided case, and it is worth checking before the next claim is filed rather than after.

What this means for a claim in preparation

Neither decision creates a new rule. Both are tribunals applying the test that was already in the legislation and the DSIT Guidelines, which is the part that should concentrate the mind: there is no commencement date to work around and no transitional comfort to rely on. Four things follow for work being done now.

State the baseline against named published sources. The baseline section should identify the publicly available knowledge the work was tested against — standards, textbooks, trade and industry guides, published papers, vendor documentation — and say why the claimed advance sits beyond it. RDM Wines lost on a handbook the Tribunal found for itself. A narrative that does not address the obvious literature in its own field invites exactly that.

Keep the baseline separate from the problem. If the trigger for the project and the state of the field are expressed in the same sentence, they are almost certainly being conflated, and Environmental Services shows where that ends. Describe the field’s prior knowledge first, as a thing that exists independently of the company’s difficulties.

Establish the competent professional on all three limbs. Qualifications, relevant experience, and how their knowledge of the field is current. A role and a number of years covers two of the three. Conference attendance, standards committee work, published papers, continuing professional development, active engagement with vendors and research groups — whatever is true — covers the third.

Capture the evidence while the work is happening. The RDM Wines list is the specification: planning correspondence, test results, internal discussion of alternatives, timesheets, contractor invoices, contract terms. A supporting letter written years later is, as that decision shows, worth very little. The project start date in particular should be the date the documents support, not the date anybody remembers.

None of this is onerous while a project is live. All of it is impossible once the project is three years finished and under enquiry.

Where advice actually changes the outcome

The instinctive reading of two lost appeals is that the companies should have taken proper advice before going to tribunal. That is true as far as it goes, but it is the wrong point in the timeline.

By the time an appeal is being contemplated, almost everything that decides it has already happened. The contemporaneous records either exist or they do not. The baseline was either considered when the narrative was written, or it is being constructed retrospectively — which is precisely what the Tribunal identified and discounted in RDM Wines. No adviser, however good, can produce a 2021 planning email in 2026, and the attempt to supply the gap with recollection and a later letter is what the Gestmin reasoning is designed to filter out.

So the advice that changes outcomes sits much earlier: at the point the projects are identified, while the work is still being done, and when the narrative and the cost schedule are being built. That is when the field’s published knowledge can still be researched and addressed, when the competent professional’s currency of knowledge can still be evidenced, when hire and lease costs can still be put in the right category, and when a project’s start date can still be documented rather than remembered.

There is a second place advice matters, and it is the one these decisions speak to most directly. Where an enquiry is already live, the useful question is not how to argue the case but what the available evidence will actually support. A realistic assessment — including, sometimes, the assessment that a position is not worth defending to a hearing — is worth more than an optimistic one. Both of these appeals were run and lost, with the costs and the time that implies, on arguments that the documents could not carry.

That is a judgement call that benefits from someone who reads these decisions as they come out and knows what tribunals are currently doing with them. It is not a judgement that improves by being made later.

The decisions

Also cited in these decisions: Tanglewood Care Services [2026] UKFTT 1137 (TC), on reading paragraph 9 of the DSIT Guidelines together with paragraph 6, and Flame Tree Publishing, on the competent professional test.

This article discusses published tribunal decisions and is general information about how the R&D relief tests are being applied. It is not advice on any particular claim, and whether a given project or cost qualifies depends on its own facts.