R&D tax incentives for manufacturers: where the qualifying work hides

Software companies claim readily because their work looks like research. Manufacturers often do not claim at all, because their R&D is done by production engineers solving problems on the line and nobody writes it up. Yet the definition of qualifying R&D in every one of the seven schemes turns on scientific or technological uncertainty addressed systematically (Singapore's definition refers to novelty or technical risk), and a factory generates that constantly.

Where to look

Process development

Getting a new product to run on existing equipment, or an existing product to run faster, cleaner or with less waste, frequently involves trials whose outcome could not be predicted. The engineering time spent on trial runs, parameter changes and measurement is a candidate.

Materials and formulations

Substituting a material or ingredient while holding performance, cost and regulatory properties constant is classic experimentation, and the failed batches are the evidence.

Tooling and fixtures

Designing tooling for a part or tolerance the supplier said could not be achieved, and iterating until it can, is development work even though the output is a jig rather than a product.

Scale-up

A process that works at bench scale and fails at production scale presents a new technical unknown. The work to close that gap is often the largest single candidate in a manufacturing claim.

Automation and integration

Integrating machines, sensors and control systems in a configuration the vendors did not design for, where the combined behaviour had to be discovered by testing.

Quality and reliability engineering

Root-cause investigations into failures where the cause was genuinely unknown and had to be found by structured experiment, as opposed to routine inspection.

What to exclude

• Commissioning equipment to the manufacturer's specification.
• Routine quality control, calibration and maintenance.
• Production runs once the process is stable, even of a new product.
• Cosmetic changes, packaging design and market-driven variants with no technical unknown.
• Efficiency gains achieved by applying known methods.

The test is the same as everywhere: could a competent engineer have known how to do it without trials? If yes, it is good engineering, not R&D.

Evidence on the shop floor

Manufacturing evidence is different from software evidence but no harder to keep:

• Trial run sheets with parameters, results and the engineer's name, dated.
• Batch records for failed or experimental batches, retained rather than discarded.
• Engineering change requests that state the problem being solved.
• Photographs and measurements of prototypes and tooling iterations.
• A project note at the start of each development stating what is unknown.
• Time captured by project for engineering staff, monthly at the latest.

Costs that are often missed

Materials consumed in qualifying experiments and the time of staff directly working on them can be eligible, but each scheme limits trial production. New Zealand, for example, excludes preproduction trial runs and tooling up from core activities, and Canada counts only the experimental part of commercial production. Utilities and consumables where the scheme allows them. Subcontracted testing and analysis, subject to each scheme's rules on contracted work. A finance team that only looks at engineering salaries will under-claim.

Getting started

Run a project identification workshop with production and engineering leads (our guide sets out the format). Expect the first list to be long and the challenge round to cut it hard. What survives is usually well worth claiming, and the evidence for it is already in the plant.

Educational material, not advice. Eligibility and eligible costs depend on the scheme and the facts.