Research and development tax relief: the records the claim rests on

Research and development tax relief reduces a company’s tax bill, or pays cash, for work that sought to resolve a genuine scientific or technological uncertainty. Most rejected claims are not rejected because the work was ineligible. They are rejected because nobody wrote down, at the time, what the uncertainty was and how it was tackled — and a narrative reconstructed two years later from memory and invoices reads exactly like a narrative reconstructed two years later.

What qualifies, roughly

• An advance sought in a field of science or technology — not in your own knowledge, and not in commercial terms. Building something new to you but routine to a competent professional in the field does not qualify.
• A technological or scientific uncertainty: at the outset, a competent professional could not say whether or how it could be done.
• Systematic work to resolve it, including the attempts that failed. Failed work is qualifying work, which surprises people.
• The activity stops qualifying at the point the uncertainty is resolved — subsequent build-out is normal development.
• Software counts, but "we used a new framework" is not an uncertainty. "We could not establish whether this could be made to perform at this scale" may be.

Research and development tax relief: what to record as you go

1. A short project record per piece of work: the uncertainty in a paragraph, the approaches tried, what was learned, and when it stopped being uncertain. Written at the time, dated.
2. Who worked on it and roughly what proportion of their time — apportionment is where most of the value sits and where most claims are weakest.
3. Costs by category as the regimes define them: staff, subcontractors, consumables, software. Tagging them in the ledger during the year is minutes; separating them afterwards is days.
4. Failed attempts, kept rather than deleted, because they are evidence that the work was genuinely uncertain.
5. The competent professional’s view — someone technically credible saying why this was not routine. That person is usually your own engineer, and their contemporaneous note is worth more than a consultant’s later prose.

The rules, rates and definitions differ substantially between jurisdictions and change often — the UK regime has been rewritten repeatedly, and the US credit works on an entirely different basis. Nothing here is tax advice. What is consistent across regimes is the evidential expectation: contemporaneous records of the uncertainty, the work and the cost apportionment.

Where claims go wrong

• The narrative describes what was built rather than what was uncertain — the single most common weakness.
• Time apportionment estimated in one sitting at year end, with round numbers that do not correspond to anything.
• Routine development included because it happened during the same project, which invites scrutiny of the whole claim rather than that part.
• Subcontractor and grant interactions ignored, where the treatment differs and can reduce or disqualify relief.
• A claim prepared entirely by an external adviser with no internal technical input, which is increasingly the pattern tax authorities look for.

Where the records live

Ettex Records holds one record per project with the uncertainty, the dated notes, the people involved and the cost categories — which turns claim preparation into an export rather than an investigation. The point is that the notes exist with dates on them before anyone thinks about a claim; a record created in the claim window is worth little regardless of what it says. Keep the cost tagging in the ledger via the chart of accounts rather than in a parallel spreadsheet. Ettex does not prepare or file claims, does not calculate relief and gives no tax advice — use an accountant or a specialist, and be wary of anyone charging a percentage who does not want to speak to your engineers.