Definition
Cosmetic uncertainty arises when a problem looks difficult because of its scale, context, or novelty to the particular business, but is solvable by applying techniques already documented in the literature or known to practitioners in the field. A competent professional encountering it would know how to proceed without needing to run experiments to determine whether a solution is achievable.
Substantive uncertainty is the opposite condition. The answer to the technical question is not readily deducible from existing knowledge, even by a skilled practitioner. The company must investigate, experiment, or develop a new approach precisely because no established path leads to the result. This is the condition the BIS Guidelines on the meaning of research and development for tax purposes describe when they refer to uncertainty that cannot be resolved without R&D activity. HMRC's guidance at CIRD81900 frames the test in similar terms: would a competent professional know, at the outset, how to achieve the intended result?
How the distinction applies in practice
The BIS Guidelines, published at gov.uk, require that the uncertainty be in science or technology, not in commercial, financial, or operational matters. A business may face genuine commercial uncertainty about whether a product will sell, but that is not R&D uncertainty. Similarly, building a product in a new market context using known techniques does not generate substantive technological uncertainty, even if the business has not done it before.
The competent professional test is the reference point. If a developer with current knowledge of the field could look at the problem and know the answer, or know reliably how to find the answer, the uncertainty is cosmetic. If even that developer would need to run experiments or prototype approaches to determine whether the result is achievable at all, the uncertainty is substantive.
Example
A software company builds a reporting dashboard for a client. Combining several APIs and writing query optimisation logic is complex but follows documented patterns. This is cosmetic uncertainty: the methods are known, the outcome is predictable, and the work is engineering rather than research.
The same company then attempts to train a machine learning model to classify unstructured text in a domain with limited labelled data and high accuracy requirements that existing published approaches do not meet. The team does not know at the outset whether the required accuracy is achievable, what architecture will work, or whether the problem is tractable with available data. That is substantive uncertainty.
Common mistakes
The most frequent error is writing up cosmetic uncertainty as though it were substantive. Phrases like "we did not know if we could integrate X and Y" describe implementation complexity, not technological unknowing. HMRC reviewers have seen this pattern widely enough that the Additional Information Form now asks for the specific uncertainty at the outset of the project, not a description of what was built.
A second mistake is conflating business risk with technological uncertainty. Launching a new product is risky. That risk does not generate an R&D claim unless the underlying technology is itself unknown. See also the routine improvement boundary, which applies where the advance sought is incremental but the method is already well-understood.
If you want to test whether your project clears this line, the eligibility checker poses the key questions in plain language.