How do I work out what was already known in my field?

Work out what a competent professional in the field could have found, or readily deduced, from publicly available knowledge at the point the project started. That is your baseline. Everything in the claim is measured against it: the advance is what goes beyond it, and the uncertainty is what a professional could not resolve. Establish it before the project, record what you looked at, and date the record.

Establishing the baseline for an R&D claim

What the baseline actually is

The guidelines define overall knowledge or capability in a field as the knowledge or capability that is publicly available, or readily deducible from publicly available knowledge or capability, by a competent professional working in the field.

Read that definition slowly, because it has three moving parts and claims fail on each of them.

Publicly available. Not what your company knew. Not what your team had access to. What is in the public domain, anywhere in the world.

Readily deducible from it. The baseline is not just what has been written down. It includes what a professional could work out from what has been written down without significant effort. This is the part that catches optimistic claims: nobody has published exactly your solution, but a competent professional would have got there in an afternoon.

By a competent professional in the field. The baseline is not what an intelligent generalist could find. It is what someone with real standing in the field could find and deduce, using the ordinary sources of that field.

What counts as publicly available

Wider than most people assume, but not free or easy to find. Publicly available knowledge includes:

published research, journals and conference proceedings, including material behind a paywall;
patents and published patent applications, including expired ones;
technical standards and specifications;
vendor and manufacturer documentation, datasheets and reference designs;
textbooks and standard reference material;
open-source code and its documentation;
knowledge that is simply common currency among practitioners in the field, whether or not anyone has written it down.

The last has no paper trail and is what your competent professional is there to speak to.

What is not publicly available: a competitor’s trade secrets, unpublished internal work, and knowledge that exists only inside organisations that have not disclosed it. If an advance has been made somewhere but the details are not in the public domain, work to reach it can still be R&D.

What a baseline search looks like

There is no prescribed method, but a defensible baseline usually has four elements.

The field, stated. Which field the project sits in, and therefore which body of knowledge is the reference. Projects that span fields need a baseline for each.
The sources actually consulted. Named, with dates. Literature, patents, standards, vendor material, and who was asked.
The conclusion. What the field could and could not do, in the specific respect the project cares about, at that date.
Who reached it. Your competent professional, by name, with a note of why they are qualified to say.

None of this needs to be long. A page, written by the right person, dated before the work started, is worth more than thirty pages written afterwards by someone else.

The baseline moves

A baseline is a statement about a moment. The field advances, and so a project that qualified in its first year may not qualify in its third — including where it was your own company that moved the field, by publishing, patenting or shipping.

This matters for multi-year claims. The right approach is to re-establish the baseline at the start of each accounting period and to be prepared to say that the later work is no longer R&D. Rolling the same technical narrative forward for three years, unchanged, is one of the most reliable ways to attract an enquiry into all three periods.

Position for accounting periods beginning before 1 April 2023

The earlier guidelines, issued in 2004 and updated in 2010, define overall knowledge or capability in the same terms. Nothing in this entry turns on the boundary, except that for those earlier periods the relevant field cannot be pure mathematics.

Worked example

Illustrative. A manufacturer wants to bond two dissimilar materials in an assembly, at a joint strength no existing adhesive achieves at the operating temperature.

Baseline element	What the record shows
Field	Adhesive science and polymer chemistry, plus joint mechanics
Sources consulted	Four adhesive manufacturers’ technical datasheets; two named standards; a patent search on the two material classes returning 31 results, 6 read in full; two review papers; an enquiry to a named supplier’s technical team, answered by email
Date of the search	Recorded, three weeks before the first laboratory work
Conclusion	Published systems reach roughly two-thirds of the required strength at temperature; no published system addresses the combination; two patents describe surface treatments that might be adapted, with no data at this temperature
Reached by	Named materials engineer, chartered, 12 years in adhesives

That last row is what makes the baseline usable. The conclusion also does something claims rarely do: it identifies the two patents that come closest, which is the strongest available evidence that the search was real. A baseline that finds nothing at all is usually a baseline that did not look.

Where claims go wrong

No baseline at all. The claim describes what was built and asserts that it was novel. There is nothing to measure the advance against, so the advance cannot be established. This is the most common structural defect in claims we review.
The baseline is the company’s own prior position. “Our previous system could only do X” describes an internal starting point. The test is what the field could do, and the two are rarely the same.
A search that was not a search. A single web search, or a statement that “no commercial product does this”, is not a baseline. Product availability is a market fact; the baseline is a knowledge fact.
Confusing not-published with not-deducible. Nobody has written down your exact configuration. That does not make it an advance if a competent professional would have derived it from what is published.
Written afterwards. A baseline produced during the claim, describing the state of knowledge two years ago, is worth much less than a contemporaneous one — and where it conflicts with anything in the project record, it does active harm.
One baseline for several projects. Different projects sit in different fields or address different questions within a field. A single baseline stretched across a portfolio does not survive a project-by-project request.