Where the science genuinely runs out
Qualifying work is where a competent agronomist, scientist or agricultural engineer could not readily deduce the answer. In agri-biotech, novel diagnostics, methane-suppressing feed additives, biopesticides or biostimulants where the mode of action or formulation stability is genuinely uncertain. In precision agriculture, developing rather than deploying machine-vision detection, sensor fusion or variable-rate algorithms. In machinery, autonomous field robots and harvesters that must navigate and manipulate in unstructured outdoor conditions. In controlled-environment and vertical farming, novel lighting spectra and climate or nutrient-delivery systems where the interactions are not deducible, or a step change in energy efficiency. And in crop and soil science, gene-editing or marker-assisted approaches with uncertain outcomes, and novel soil-amendment technologies.
The work that feels like R&D but is not
Most farming activity is outside the definition, however skilled. Routine agronomy, rotation, fertiliser timing and planting density to established practice. On-farm variety trials that evaluate known varieties. Adopting and configuring existing precision-agriculture kit. Seasonal or weather variation dressed as an experiment. Standard selective breeding producing an expected improvement. And certification, organic conversion or assurance-scheme compliance. Doing something established in the industry but new to your farm is not an advance.
Where the boundary bites
Two things settle most agriculture claims. The first is the “trial and error on the farm” trap: iterative work qualifies only where it is systematic investigation of a genuine technological uncertainty, not optimisation of known variables, so the actual uncertainty and the competent professional have to be pinned down. The second is funding: agriculture is heavily grant-supported, and while grants no longer reduce a claim for periods from April 2024, the old restriction still bites for earlier periods and interacts with the enhanced route, so we check the period first.
Sector pitfalls we would check first
- Optimisation is not R&D. Adjusting known variables to improve a yield is farming; resolving a genuine technological uncertainty is R&D. The claim has to name which it is.
- The grant period. Grant funding no longer reduces the claim for periods from April 2024, but the period has to be confirmed, and grant-funded costs must not be double-counted.
- Contracted-out work. Where research runs through co-operatives, contract growers, levy bodies or institutes, the party that intended and contracted the R&D claims, not necessarily the one doing the work. The subcontractor rules decide it.
- Produce sold from a trial. Inputs consumed in a trial can qualify, but where the resulting crop or livestock is sold in the ordinary course, the cost is restricted; capital such as glasshouses and machinery sits outside the revenue claim. The rules are on qualifying costs.
Frequently asked questions
We trial different varieties on the farm each year. Does that qualify?
Generally no. Evaluating which known varieties perform best in your conditions is commercial evaluation, not R&D, and natural year-to-year weather variation is not a systematic experiment. It qualifies only where you were resolving a genuine technological uncertainty a competent agronomist or scientist could not readily resolve.
We adopted precision-farming kit and it transformed the farm. Is that R&D?
Not the adoption. Deploying and configuring existing GPS, drone or sensor technology as intended is use, not advance. Developing genuinely new detection, sensing or control technology, where the technical approach was uncertain, is different and can qualify.
Is "new to our farm" enough?
No. The advance has to be in the field's overall knowledge or capability, not new to your business. However novel to you, if it was already achievable in the industry there is no R&D.
We have a Farming Innovation Programme grant. Can we still claim?
Yes, for accounting periods beginning on or after 1 April 2024: the merged scheme dropped the restriction that reduced grant-funded R&D. For earlier periods still in the amendment window the old rules apply, so the period matters, and grant-funded costs must not be double-counted.