Manufacturing is one of the biggest claiming sectors. Production runs are not R&D.

Work to develop or improve a product, material or process where the outcome was uncertain and a competent professional could not readily have deduced the answer. That might be a new material mix, a process that has to hit a tolerance nobody has hit before, or a method of removing a contaminant. Making more of an existing product the same way is production, not R&D.

It can, if the improvement is an appreciable one that required resolving technological uncertainty. The Government's guidelines include appreciable improvements in the definition of an advance. Tidying a workflow, buying a better machine and installing it as the supplier intended is not. The question is whether you had to work out something that was not known or readily available.

Yes, where the run is part of the R&D, for example testing whether a new process holds at line speed. The cost of materials consumed and the staff time on the trial can qualify. A run that is just early production for sale does not. We look for a test plan, a measured result and a decision taken because of it.

Materials used up in R&D are consumables and can qualify, along with items such as fuel, chemicals, power and water. Whether the cost of a prototype as an asset can be claimed is a different question, because capital expenditure is excluded. We split the spend between items consumed in testing and anything that is kept as an asset.

The Government's guidelines deal with the start and end of R&D. In broad terms, R&D ends when the uncertainty has been resolved or work to resolve it stops. After that, runs are production. We set the end date with you using the project records, and costs after that date are not claimed under R&D.

If the batches were consumed in R&D testing, the materials in them can be consumables. A batch that fails during a trial of an uncertain process is evidence as well as a cost, because it shows the problem was real. Scrap from normal production, or from a process that is already proven, is not R&D.

It depends on how the tooling is treated. Capital expenditure is not claimable under R&D tax relief, so tooling that is an asset is generally out. The staff time spent solving a real technical problem in designing it, and materials consumed testing it, can qualify. A separate relief for capital spending on R&D may help with the asset itself.

Sometimes. Buying and installing a standard robot is not R&D. Getting a robot or control system to do something that has not been done, such as handling a variable material at speed, can be if the engineers faced real uncertainty. The records need to show what could not be predicted and how it was resolved.

Adopting existing technology is not an advance, because the Government's guidelines say using technology is not an advance in itself. The work can qualify where you had to resolve uncertainty to make it function in your setting, for example because the data could not be captured by any known method. Installation, licences and training alone are not R&D.

Payments for contracted-out R&D can qualify if you decided the R&D needed doing and can evidence that decision. For an unconnected laboratory, 65% of the payment is the qualifying amount. For periods beginning on or after 1 April 2024, the work must be done in the UK or meet the overseas exception. Routine compliance testing is a different matter.

They can be, for the time they spend directly on R&D. Operators who run a trial under the engineers' direction, and technicians who make adjustments and record results, are doing R&D work during that time. The same people on normal production are not. We rely on records or a reasoned method to apportion time.

It can, if the work resolves technological uncertainty. A sustainability aim is not itself the test. Replacing a material with a recycled or bio-based one often raises real problems of performance and processing. Our recycled plastics case study is an example of regulation and material shortages forcing a manufacturer to solve new technical problems.

With test plans, trial logs, measurements, samples and photographs, internal reports and emails in which engineers discussed why something was not working. Where a laboratory ran tests, keep its reports. The aim is to show what you did not know, how you set about finding out and what you found, preferably in records made at the time.

Testing a finished product to demonstrate that it meets a standard is usually routine. Work to find out whether a product can be made to meet a standard, where the answer was uncertain, may qualify. The line is between proving compliance and resolving a technical problem on the way to it. We look at what the testing was actually for.

They can, but bespoke does not mean R&D. A one-off built using established methods is routine. A product that the customer needs and nobody knows how to make, which you set out to develop, may be R&D. If the customer pays for the development, check who decided the R&D needed doing before assuming you can claim.