Software companies remain some of the UK’s most frequent claimants of R&D tax relief, representing over 40% of all R&D Tax Credit claims. From AI-powered applications to cloud-native platforms and complex SaaS solutions, innovation continues to drive growth across the technology sector.

However, following recent reforms to the UK’s R&D tax relief schemes and increased scrutiny from HMRC, it is more important than ever to understand what genuinely qualifies as Research and Development (R&D).

In this guide, we explain what software companies can claim for in 2026, where HMRC draws the line, and common mistakes that could put an R&D claim at risk.

Are UK SMEs Being Shut Out of R&D Tax Credits Under the New Merged Scheme?

What Counts as Software R&D?

Many businesses assume that creating software automatically qualifies for R&D tax relief. Unfortunately, that is not the case.

For a project to qualify, it must seek to achieve an advance in science or technology by resolving Technological Uncertainties that could not easily be solved by a Competent Professional.

In simple terms, your developers must be solving genuinely difficult technical challenges—not simply building software using existing methods or technologies.

Examples of qualifying activities include:

  • Designing new software architectures
  • Developing novel algorithms
  • Improving system performance beyond current industry standards
  • Solving complex scalability or security challenges
  • Creating innovative integrations where no established solution exists

Routine software development, feature updates, bug fixes, and cosmetic improvements generally do not qualify.

Does AI Development Qualify?

Artificial Intelligence (AI) remains one of the fastest-growing areas for R&D claims in 2026, but not every AI project qualifies.

Many companies now integrate existing AI models or APIs into their products. While this can deliver commercial value, simply implementing publicly available AI tools is unlikely to constitute R&D.

Projects are more likely to qualify where they involve:

  • Developing proprietary machine learning models
  • Creating new training methodologies
  • Improving model accuracy through novel approaches
  • Solving complex optimisation problems
  • Building unique natural language processing or computer vision techniques
  • Overcoming challenges relating to inference speed, scalability, or data efficiency

For example, creating a bespoke fraud detection engine capable of identifying previously undetectable financial patterns may qualify if significant Technological Uncertainties had to be overcome.

Conversely, integrating an existing large language model into a customer support chatbot without advancing the underlying technology would generally not qualify.

SaaS Platforms: What Can Be Claimed?

Software-as-a-Service (SaaS) businesses frequently undertake qualifying R&D, particularly during platform development.

Eligible work may include:

  • Designing highly scalable multi-tenant architecture
  • Developing complex distributed systems
  • Building innovative data processing pipelines
  • Creating new methods of synchronising large datasets
  • Improving resilience and fault tolerance
  • Solving low-latency performance challenges

Many successful claims arise from solving infrastructure-level problems rather than developing customer-facing features.

The key question is always:

Was there genuine Technological Uncertainty that required experimentation to resolve?

If the answer is yes, there is a strong possibility that at least part of the project may qualify.

Cloud Engineering and Infrastructure Innovation

As more organisations move to cloud-native environments, cloud engineering has become another significant area for R&D claims.

Qualifying activities may include:

  • Developing automated deployment frameworks
  • Creating new container orchestration techniques
  • Designing resilient microservices architectures
  • Improving distributed database performance
  • Developing innovative security controls
  • Solving infrastructure scaling challenges

Simply migrating an application from on-premises servers to a cloud provider is unlikely to qualify if established migration techniques are used.

However, overcoming novel engineering challenges during migration or creating new methods of managing cloud infrastructure could potentially qualify.

What Doesn’t Qualify in 2026?

HMRC continues to scrutinise software claims carefully, particularly where projects appear to involve routine development. We understand HMRC are also using AI to judge R&D project eligibility.

Examples that are generally not eligible include:

  • Standard website development
  • Mobile app development using conventional techniques
  • Cosmetic UI or UX improvements
  • Routine software maintenance
  • Bug fixing
  • Performance improvements achieved using known methods
  • Installing or configuring third-party software
  • Using existing AI tools without advancing the technology
  • Migrating systems using standard cloud practices

Commercial innovation alone is not sufficient.

A product can be new to your business—or even new to the market—and still fail to qualify if it does not involve overcoming genuine Technological Uncertainty.

Keeping the Right Evidence

One of the biggest reasons software claims are challenged is poor documentation.

HMRC increasingly expects businesses to demonstrate:

  • The technological challenges encountered
  • Why existing solutions were insufficient
  • The experimentation undertaken
  • Failed approaches and iterations
  • Technical decisions made throughout the project
  • The staff involved and the time they spent on qualifying activities

Maintaining detailed project documentation throughout development can significantly strengthen a claim and reduce the likelihood of an Enquiry.

Common Mistakes Software Companies Make

A lot of businesses unintentionally overclaim by including all software development costs.

Some common errors include:

  • Claiming entire projects instead of qualifying elements
  • Including project management or marketing activities
  • Treating commercial uncertainty as Technological Uncertainty
  • Claiming routine coding work
  • Failing to document technical challenges
  • Assuming AI-related work automatically qualifies

RDP understands the nuances of preparing R&D Tax Credit claims, which result in accurate and better-supported submissions.

Final Thoughts

Software companies continue to be among the strongest candidates for UK R&D tax relief in 2026. Whether you are developing AI solutions, building scalable SaaS platforms, or solving complex cloud engineering challenges, valuable relief may be available if your work advances technology through genuine innovation.

However, the rules are more rigorously enforced than ever. Success depends not only on undertaking qualifying R&D but also on clearly demonstrating the Technological Uncertainties you faced, and the work carried out to overcome them.

RDP has been assisting companies in all industries with their R&D Tax Credit claims for over 20 years. In preparing R&D Tax Credit claims for our clients, our objectives are:

  • Ensuring all eligible projects, activities and costs are claimed. RDP’s proprietary Innovation Connection Program (“ICP”) ensures this happens
  • RDP backs up its work by providing full support should HMRC compliance of the claim occur. We have an exceedingly high success rate on compliance checks.
  • Our ICP also ensures the R&D Tax Credit is fully documented and supported

If you have any questions on our service, please do not hesitate to contact Jenni at [email protected].