Time tracking for R&D tax credit claims
R&D relief schemes turn on two questions, and only one of them is a software question. Whether your project qualifies is for your adviser. Whether you can show how many hours went into it, on the days they went into it, is a records question — and most companies answer it with a spreadsheet filled in months later from memory, which is the part of a claim an enquiry pulls at first. ActivityPulse attributes engineering time to projects as the work happens, so the staff-cost apportionment rests on what was observed rather than on what was recalled.
Why contemporaneous beats reconstructed
The distinction is old, unglamorous, and the single thing that separates a claim that is tiresome to defend from one that is not.
A contemporaneous record is one made at the time the work was done. A reconstruction is one made later, usually by someone looking at a calendar, a ticket tracker and a deadline. Tax authorities across the major schemes treat the first as stronger evidence, and the reason is not bureaucratic fussiness: a record made on a Tuesday in June was not made with knowledge of the size of the claim.
Reconstructions are not forbidden, and plenty of claims are built on them. But they are where an enquiry starts, and answering one costs the time of exactly the people whose time you were claiming for. The asymmetry is what makes this worth fixing in advance: capturing the record as you go costs nothing extra, and reconstructing it later costs weeks precisely when you can least afford them.
The practical failure is subtler than a missing timesheet. Engineers do not split their week into clean blocks — a quarter is qualifying work, customer escalations, a migration, and three days of something nobody named. A spreadsheet resolves that into a round percentage. Continuous attribution does not have to.
What this actually gives you
Records, in the shape an adviser can work from.
- Hours attributed to a named project on the day they were worked, not apportioned afterwards.
- The split between qualifying work and everything else, measured continuously rather than estimated in blocks.
- Corrections retained as marked, audited adjustments, so the original record survives the amendment.
- Per-person, per-project, per-day detail that an adviser can aggregate the way their scheme requires.
- An export, so the evidence leaves the product in a form a finance team can reconcile.
What it does not do
Said plainly, because the vendors who blur this line are the reason the question gets asked carefully.
- It does not decide whether a project qualifies for relief. That is a technical and legal judgement about the work.
- It does not calculate a claim, a rate, or an amount.
- It does not produce a scheme-specific form, and no tax authority has approved or endorsed it.
- It cannot evidence a period that ended before it was installed.
The rollout conversation is easier here
Most monitoring deployments struggle because the purpose is vague and the scope feels open-ended. This one is neither. The company is claiming relief on work its engineers did, the records exist to support that claim, and the thing being recorded is which project someone worked on. That is a purpose a team can evaluate and agree to, which is a materially different conversation from being told a tool is being installed.
It still has to survive the same scrutiny as any other rollout. In several European jurisdictions monitoring requires consultation with a works council or employee representatives before it starts, regardless of how reasonable the purpose is — see the per-country guides, and the page written for whoever owns that step.
What the design guarantees
- manual-adjustments-audited
- A manually adjusted timesheet is marked as adjusted and audited; captured data is never silently overwritten.
- no-double-count
- An hour is never counted twice: each activity segment carries a UUIDv7 the server upserts on, so a retried sync adds nothing.
- no-keystroke-logging
- Keystroke content is never captured — there is no code path that records what a person types.
- no-productivity-score
- ActivityPulse reports what happened and does not grade the person it happened to — there is no productivity score.
Frequently asked questions
What are contemporaneous records, and why do they matter for an R&D claim?
A contemporaneous record is one created at the time the work was done, rather than reconstructed afterwards. Tax authorities across the major schemes treat them as stronger evidence than a retrospective estimate, for the straightforward reason that a record made on the day was not made with the size of the claim in mind. A reconstruction is not automatically rejected, but it is the thing an enquiry probes first, and defending it consumes the time of the people who did the work.
Does ActivityPulse make our R&D claim succeed?
No, and treat any software that says otherwise with suspicion. Whether a project qualifies turns on the technical and scientific facts of the work and on the scheme's rules, neither of which a time-tracking tool can assess. What this product does is narrower and genuinely useful: it records which project a person's hours went to, on the day they went there, so the apportionment underlying the staff-cost part of a claim rests on observed time rather than on a recollection.
Our engineers split their week between qualifying work and everything else. How is that handled?
That split is the whole difficulty, and it is why round numbers in a spreadsheet are hard to defend — an engineer who spent 60% of the quarter on the new pipeline and 40% on customer support did not spend it in neat blocks. Time is attributed to a project continuously as the person works, so the proportion is a measurement rather than an estimate. How that proportion should then be applied is your adviser's call.
Can we correct the record if a project was attributed wrongly?
Yes, and the way corrections behave is deliberate. A manual adjustment is supported, marked as an adjustment, and audited — it does not silently overwrite what was captured. A record that can be rewritten without trace is worth very little as evidence, so the original survives the correction and both are visible.
We are claiming for a period that has already ended. Does this help?
Not for that period, and it would be dishonest to suggest otherwise. Records cannot be created retrospectively — that is the entire point of the word contemporaneous. It helps for the period you are in now, which in practice means the claim after next.
Does this work for schemes outside the UK?
The product records time against projects; it does not implement any scheme's rules. The major regimes — UK R&D relief, the US research credit, Canadian SR&ED and their equivalents — differ substantially in what qualifies and in what they expect to see, but they have in common that staff time is usually the largest cost line and the one most often questioned. Ask your adviser what evidence your scheme expects, then check whether this produces it.
Will our engineers accept being tracked for this?
It is an easier conversation than most monitoring rollouts, because the purpose is legible and finite: the company is claiming relief on work they did, and the records support the claim. It helps that keystroke content is never captured, screen capture is off by default and behind a compliance review, no productivity score is produced for anyone, and the person can see what is being collected on their own machine and pause capture.
Not tax advice
This page is about evidencing staff time. It is not tax advice, and nothing here is a view on whether a particular project, cost or company qualifies for relief under any scheme — that is a question for a qualified adviser or accountant who knows your facts. Schemes, rates and record-keeping expectations differ by country and change frequently.
See where the hours actually went
Free for up to three people. Windows and macOS. No keystroke content on any plan, and screenshots off unless an administrator turns them on.