If your practice has been preparing well-evidenced R&D claims for years and nothing has gone wrong, keep doing it. What follows is what's actually different now, and whether it applies to you.
What's actually changed
A formal registration obligation, tied to interacting with HMRC
Advisers who interact with HMRC on a client's behalf now fall under a formal registration requirement under the Finance Act 2026, Part 7 (Schedule 20 sets out the exemptions to it, not the requirement itself). Separately, HMRC's compliance activity on R&D claims generally has stepped up in recent years. Neither of those facts says careful work is now unwelcome. What they add up to is a lower tolerance for thin work, and a specific new compliance step attached to the act of representing a client to HMRC on this, regardless of how much or how little of your practice's revenue R&D work makes up.
The first real question
Is the technical-narrative skill actually there, separate from your tax competence
General tax and accounts work and R&D technical-narrative writing aren't the same skill, even though the same person often ends up doing both. A claim's technical narrative has to describe a genuine, specific uncertainty (what wasn't obvious in advance, what was tried, and how it was resolved) in a way that couldn't just as easily describe any other project. That's closer to specialist technical writing than to a tax computation. A practice that's been doing this well for years, with real project detail and not boilerplate, almost certainly already has this skill, whether or not anyone in the firm has named it that way. A practice picking up its first R&D claim because a client asked is starting from a different position, and that's worth being honest about before taking it on as routine work.
In practice, this is where claims fail, not at the notification deadline: a technically strong piece of underlying work can still produce a weak claim if the written narrative reads as generic. That's a writing-and-evidencing risk, not a knowledge-of-tax-law risk, and it's the one that's easiest for a general practice to underestimate, because it looks like normal client-file work rather than a distinct discipline.
The second real question
Do you want the registration obligation for a small part of your practice
The Finance Act 2026 registration requirement doesn't scale with volume: a firm preparing one R&D claim a year for a long-standing client takes on the same obligation as a firm doing it regularly. That's not a reason to refer by default. Some practices will decide the obligation is worth it precisely because the client relationship and the fee both matter to them. But it is a real cost that a firm doing R&D as an occasional bolt-on should weigh against what the work is actually worth to the practice, rather than treat as a formality to file away.
What this doesn't mean
It isn't a verdict that every general practice should refer
It doesn't mean the work has become more likely to be challenged just because scrutiny generally has increased. A well-evidenced claim from a competent adviser is still a well-evidenced claim. And it doesn't mean every general practice should refer: plenty of firms have the skill, the appetite, and now the registration to keep doing this properly. The point of this page is to separate the two decisions (skill and obligation) rather than let a single reaction to "more scrutiny" answer both at once.
Decision helper
Alternatives and limitations
This page assumes you already do R&D work in-house. If you're weighing the decision fresh for a client who's just asked for the first time, start at R&D tax relief for accountants instead. And this isn't a substitute for taking your own advice on the registration requirement itself. Schedule 20's detail is a legal question for your own professional body or legal adviser, not something a page like this can settle for you.
