GOL’s R&D Tax Credit Claim vs HMRC’s Claim Denial:What were the key arguments in GOL’s successful appeal?

If you don't have time to read the 24 page report, on this FIRST-TIER TRIBUNAL TAX CHAMBER appeal, about whether the Small & Medium Enterprise claimant, Get Onbord Ltd (GOL), had met the tests for undertaking qualifying “research and development” in line with the BEIS Guidelines, here is my summary of it (Appeal reference: TC/2022/13281 CORPORATION TAX, Judgment date: 09 July 2024). The appeal was against an HMRC Research & Development corporation tax compliance check which concluded that the R&D claimed did not qualify.

Note: HMRC has a period of time to appeal, so at the time of writing (5^th August 2024) it could still appeal this case.

My summary includes the following sections:

1. Overview of the Get Onbord Ltd (GOL) R&D Project
2. HMRC Arguments Against the GOL Project Being Classified as R&D
3. Rebuttals by GOL
4. My Take on the Case

1. Overview of the Get Onbord Ltd (GOL) R&D Project

Objective: The Company aimed to develop the ONBORD system, an automated, AI-enabled on boarding solution for KYC (Know Your Customer) analysis. This system was intended to significantly improve the technology associated with AI for financial services organisations, such as banks.

Initial Investigation:

• The Company conducted a state-of-the-art review of AI use in KYC for financial services.
• It determined there was no existing publicly available or readily deducible technology to meet the project’s objectives.

Key Development Goals:

1. Algorithm Development: Create algorithms to automate the translation of risk policies into on-boarding processes. Address the challenge of data normalisation and consistent identification across various sources to avoid false positives in compliance checks.
2. Data Integration and Processing: Integrate and normalise data from multiple sources using AI for validation and auto-population. Develop methods to retrieve necessary data efficiently, avoiding upfront data calls that slow down the KYC process.
3. Client Journey Optimisation: Optimise the on-boarding process to reduce the time from client sign-up to completion of KYC checks.
4. API Development: Develop an API to enable seamless data transfer into and out of various client systems, ensuring security and adaptability.

Technological Challenges:

1. Uncertainty over developing necessary algorithms and ensuring efficient data normalisation.
2. Challenges in optimising data processing to speed up the client journey.
3. Difficulties in integrating multiple data sources and developing AI methods for on-demand data retrieval.
4. The need (how) to create a secure, adaptable API for data transfer.

Innovation and Technological Advancement:

• The project involved significant experimentation and development cycles.
• Development of new compliance flags and warnings using big data sources.
• Creation of multithreaded applications for scalable compliance checks.
• Iterative development of algorithms for shareholder identification and address validation.
• Use of machine learning models, including decision trees, for fraud, compliance, and credit checks.
• Integration of existing technologies (e.g., translation tools) with newly written code to enhance functionality.
• Development of a system to automate complex human processes, such as KYC verification and risk profiling, using AI.

Technology Tools Used by GOL

1. Application Programming Interfaces (APIs): GOL integrated APIs as part of their system to connect and utilise various software components effectively.
2. Artificial Intelligence (AI): AI was employed in several aspects of the system, including automation and enhancing the decision-making process.
3. Algorithms: Custom algorithms were developed to analyse data, make decisions, and perform other automated functions within the system.
4. Data Manipulation: Techniques for managing and processing large datasets were a crucial part of the system's functionality.
5. Database Design: GOL designed databases to store, retrieve, and manage the data required for the KYC processes.
6. Python Libraries: Open-source Python libraries were used extensively to build various parts of the system, leveraging existing tools to develop new functionalities.
7. Translation Functionality (ChatGPT): ChatGPT was integrated to provide translation services within the system, aiding in multilingual support.
8. GitHub: GOL utilised GitHub for storing and managing their codebase, enabling version control and collaborative development.
9. Automation Tools: The system included various automation tools to replicate and enhance manual KYC processes.
10. Data Analysis Tools: Tools and techniques for data analysis were a significant component, enabling the system to process and interpret vast amounts of data efficiently.

These tools collectively enabled GOL to create an automated, efficient, and scalable KYC solution, integrating advanced technologies to replicate and enhance traditional manual processes.

Outcome:

• The ONBORD project aimed to create a holistic, fast, reliable, and automated KYC solution not currently available in the market.
• Significant amounts of new code were written, establishing capabilities that did not previously exist.
• The project sought to resolve technological uncertainties and achieve appreciable improvements in automated on boarding processes for financial services.

2. HMRC Arguments Against the GOL Project Being Classified as R&D

1. HMRC's Assertion of Lack of Technological Advance: HMRC argued that the use of APIs, AI, algorithms, data manipulation, and database design might have been novel and difficult but did not constitute an advance in any technology.
2. Perception of Routine Work: HMRC viewed the activities undertaken by GOL as routine work that did not qualify as R&D. This includes the integration of existing technologies and the development of code using existing libraries and tools.
3. Existence of Similar Products: HMRC suggested that similar KYC solutions already existed in the market. They argued that GOL's system was not unique enough to be considered an R&D project.
4. Lack of Protection and Novelty: The argument was made that GOL had not sought to protect its software, indicating a lack of novel technological advancement. The use of GitHub for storing code, which is not open source, was also mentioned to support this point.
5. Use of Existing Technologies: The integration of existing technologies, such as ChatGPT for translation functions, was used to argue that GOL's project did not involve sufficient innovation. It was claimed that the innovation lay more in the aggregation and utilisation of these technologies rather than creating something fundamentally new.
6. Internal Advance: There was a suggestion that the project represented an internal advance for GOL rather than a broader technological advancement. This implies that while the project may improve GOL’s internal processes, it does not necessarily contribute to the wider field of technology.
7. Data Analysis and Processing: HMRC categorised the project as primarily involving data analysis or processing, which they did not consider to be R&D. They argued that the project's core activities were standard data manipulation tasks rather than innovative research and development.
8. Comparison with Manual Processes: HMRC acknowledged that GOL’s system aimed to replicate manual KYC processes using automation and AI. However, they argued that automating an existing manual process did not amount to a technological advance sufficient to qualify as R&D.

3. Rebuttals by GOL

In response, GOL argued that:

• Their work involved significant innovation, particularly in the integration and enhancement of existing technologies to create a unique, automated KYC solution.
• The massive amount of new code written and the iterative development process indicated substantial R&D efforts.
• The development of a holistic onboarding solution that automated complex human processes and improved efficiency was not routine and constituted a technological advance.
• The absence of similar products in the market, as confirmed by industry professionals, supported their claim of innovation.
• The use of existing technologies and open-source materials did not preclude the project from being R&D, as innovation often involves building upon and integrating existing tools.

4. My Take on the Case

At RandDTax, we have seen the arguments made by HMRC in this case mirrored in many compliance check cases and we have always rebuffed them. It often appears that HMRC is taking a ‘boilerplate' approach, deploying the same arguments across different cases with little effort to genuinely investigate the specific facts of each claim. This approach often shows a disregard for, or misunderstanding of, the correct application of the BEIS guidelines.

Regarding technology projects where the R&D case must be made at the technology level rather than describing the functionality being created, I think it shows how intertwined the two things are. Therefore HMRC has to pay some heed to what the company was trying to achieve at that level, as well as what technological R&D is involved in achieving it.

Get in touch for help with a compliance check or for support with making an R&D tax credit claim.

Article by Linda Eziquiel, Regional Director, RandDTax

For assistance with your R&D Tax Credits or with R&D Compliance Checks – get in touch.