The SR&ED (Scientific Research & Experimental Development) program has existed in Canada since before the invention of the personal computer. With the evolution of the cloud and computing, the tools available to support the SR&ED program have proliferated. The only problem is that most of the software offered by SR&ED consultants is of limited utility to clients.
Let’s dig into the types of software that are available and the evolution of the technologies. Most importantly, let’s understand what is valuable for clients.
SR&ED Consulting Software Goals
The overarching goal of SR&ED software is to support the submission of SR&ED projects to the CRA (Canada Revenue Agency). Different types of software have different focus areas:
Time Tracking
The more accurately time is tracked the more defensible a SR&ED claim is. However, time tracking can be extremely tricky. Let’s take a straightforward example of a biotech lab looking for a cure for cancer. Time tracking is relatively simple because the laboratory is only involved in experiments, detailed notes are taken, and time is accurately allocated to each step of the process. All that is required is to add up the time that all the scientists spent conducting their experiments.
Let’s look at another example of a software development team. The first challenge is that very few developers track each hour of their time. However, the bigger issue is that developers typically are not SR&ED experts so they are unable to allocate SR&ED-eligible time vs. time that does not qualify for SR&ED. These issues compound when developers don’t track their time for each development task and they are not interested in determining what is SR&ED eligible.
For larger companies time tracking becomes more important. Not only does management want to analyze the efficiency of their teams, but the CRA has higher standards for what they expect to be tracked. Staff typically have more specialized skill sets and specific projects are more easily tracked.
Document Collection
Contemporaneous document collection is a key component of filing and defending an SR&ED claim. A document may be a photo of a whiteboard, a PDF with the goals of the project or manufacturing logbooks. The CRA wants to know if there is evidence to support the narrative in the SR&ED project description.
Wouldn’t it be great if all the relevant documents were automatically organized by SR&ED project so that if the CRA asks for proof, the documents can be retrieved with ease? There are hundreds of systems to retain documents from structured databases to Google Drive.
Project Identification
One of the most challenging and important aspects of filing a quality SR&ED claim is identifying projects at the “right level.” What this practically means is that one of the key skills of an SR&ED consultant is to determine when a technological uncertainty is too small and when it is too broad. For instance, if a company was looking to increase the rendering performance of its mobile application, the SR&ED project would most likely not be “Mobile Application Development Optimization.” This topic is far too broad and may encompass a number of technical uncertainties. Identifying the projects correctly is key creating too many projects can take a considerable amount of time and having too few projects can leave the door open to a lack of specificity and the CRA denying the entire claim
Financial Data
The primary financial information needed to prepare an SR&ED claim is employee payroll, contractor payments and materials payments. The majority of expenses associated with SR&ED claims are payroll data. Therefore, one must know the name of the employee and details such as their start and end dates, base salary, bonuses and even details like vacation days can be important.
Determining which contractor invoices qualify for SR&ED can also take considerable effort. One may review the contractor agreement, and determine which portion of a specific invoice qualifies for SR&ED for a specific project and a number of other checks. It is important to check that the work the contractor did that was SR&ED eligible was undertaken in Canada. When the contractor pays for their services can also impact the inclusion of the expense in the SR&ED claim.
Materials that were transformed or destroyed in the process of experimentation may also be included in the SR&ED claim. Tracking the experimental materials vs. production materials can be tricky. In addition, if the materials were transformed there may be value in them and as a result, the entire cost of the materials may not qualify for SR&ED.
Now that we know all the components of an SR&ED claim, let’s look at the technological evolution of software provided by SR&ED consultants and in-house development teams.
1990 to 2000: The Spreadsheet Era
Back in the days when Lotus 123 and Microsoft Excel were leading the spreadsheet charge, SR&ED consulting firms developed spreadsheets for consultants to use as an internal tool. The spreadsheet would have tabs for contractors, materials and employee payroll. Almost all spreadsheets have a cost matrix. This is a table that would note the proportion of time an employee spends on SR&ED projects and multiply it by their salary to calculate the total labour expenses. It also allowed the consultants to effectively estimate the claim size and impress customers with the speed at which the calculations can be made. Laptops were just becoming mainstream so consultants could bring these powerful spreadsheets with them to create the SR&ED claim with their customers.
Every consultant had their own version of this spreadsheet. The only issue is that it did not provide the consultant with a competitive advantage. In fact, sometimes it just puts more work on their clients. Some larger accounting firms saw this as a great opportunity to send their clients a pretty-looking spreadsheet, ask them to fill in all the details and send it back to them. Shockingly this still happens in 2024. We see Deloitte, EY, PWC, KPMG and others send spreadsheet templates to companies asking for project descriptions, technical information and other SR&ED data. They ask for the most prized R&D innovations to be sent via email. There could not be a more insecure way of sharing sensitive R&D and payroll data.
2000 to 2010: The Database Era
While spreadsheets work very well for complex calculations, SR&ED consultants wanted a system that clients could access. The internet was super popular, SQL was around so many companies moved the Excel spreadsheet to an app that was available to clients via the internet. The software typically had two main features. It could track time and it could record potential SR&ED projects. Time tracking was sold as a feature that provided companies with greater claim defensibility. However, very few companies use time tracking for two main reasons. The first reason was that time tracking was the best way to minimize the size of a claim for all claims outside of manufacturing and laboratory-based claims. The people who were tracking their time did not know what was and what was not SR&ED eligible. Therefore, they would only track a portion of their time towards the project. The second massive challenge was that professionals hate time tracking. To get a scientist or engineer to accurately track their time is a challenge.
The result of poor time tracking was that the systems were rarely used by the consultant to prepare the SR&ED claim. Client non-compliance and a desire to maximize claims resulted in all but a small handful of companies using a time-tracking tool provided by their consultant.
Clients could also add potential SR&ED projects to their SR&ED software. They could jot down some notes which would provide the consultant with a background of what the technical challenges were. This software feature sometimes helped consultants but rarely helped clients. The consequence of the feature was to shift the workload to the client from the consultant. As can be expected, most clients simply did not use this feature.
While almost all major SR&ED consulting practices developed software tools, almost no one uses them. There was insufficient value to the client, and even if provided for free to the client, it resulted in more work for the client with the potential of a more negative outcome.
SR&ED consultants also felt that a system that would track which claims were submitted, the estimated SR&ED value throughout the year and a way to track projects would be helpful. While these are all helpful tools, they are more utility to the SR&ED consultant and provide some information to the client, but did not make a material impact on their SR&ED claim or process.
2010 to 2020: The API Era
Companies such as Atlassian and other project management and software management companies started to take off in the 2010s. Not only was everything being digitized but the stock market was in love with SaaS companies. Everyone was developing software to accelerate their business growth, reduce costs, and bring new products to market. With treasure troves of data in project and software development platforms, SR&ED consultants looked at ways to mine the data.
The idea is relatively simple. Let’s API to the key software tools that a company uses to extract meaningful data for SR&ED. The typical software tools include accounting systems, payroll systems, project management, and code repositories. By transferring the data from these systems to an SR&ED tool, companies would avoid needing to manually provide this information to their SR&ED consultant.
The biggest benefit is that when the CRA reviews the SR&ED claim the company can point to specific logs that demonstrate work on the technical uncertainty. The second biggest benefit is that basic information such as employees’ names can be automatically uploaded to the SR&ED software. In theory, this sounds fantastic but practically there are a number of challenges which limit the utility of this approach. There are also risks associated with connecting to accounting or payroll systems that companies do not want to expose themselves to. Additionally, the greatest drawback is that the project planning or code repository is littered with hundreds or thousands of entries making it difficult to determine what may qualify for SR&ED. Therefore, even if the company is able to self-identify what challenges most likely quality for SR&ED, there is so much noise in the system, that it is not possible for machine learning to effectively identify SR&ED projects that may not have been identified by the company.
2020 to 2024: The Reporting Era
With the proliferation of SR&ED companies also came some beautiful reporting. Charts and graphs representing SR&ED expenditures and potential refunds were eye candy to executives. Bringing together multiple sources of information and being able to consolidate, manipulate and report on the outcome is beneficial to companies. There are a number of SR&ED tools that have a very nice user interface and produce appealing charts. The question is who is looking at these and why? There is really no one interested in these charts because it is not an objective of most companies. The objective is to deliver great products or experiences to clients. It is rarely an objective to undertake R&D in order to file an SR&ED claim. Therefore, while the reporting is nice to have, once again, it is of limited practical utility to anyone.
We’ve seen artificial intelligence (AI) being bantered around with many tools. It seems that everywhere you turn there is an AI-powered button to help you get your job done better. In SR&ED consulting we see this a lot also. Boast.ai is probably the prime example. They claim their platform: “Using advanced machine learning and artificial intelligence, we analyze your R&D’s digital footprint, ensuring every minute and dollar spent on eligible projects is accounted for.” I asked ChatGPT to define AI and received this answer:
Artificial intelligence (AI) is the simulation of human intelligence in machines that are designed to think, learn, and solve problems. AI systems use algorithms and data to recognize patterns, make decisions, and perform tasks typically requiring human cognition, such as speech recognition, decision-making, and visual perception. It ranges from simple automation to advanced machine learning and neural networks that improve performance over time.
The interesting thing about what consultants are offering is that there is unlikely any AI involved. Any consultant that requires expert consultants who interview clients and prepare SR&ED claims, in the same way that has been done for 4 decades is probably not using AI. Sure they can get a bit of data via scraping the code repository or project planning database but this is not AI.
2025 and beyond: The AI Era
Like it or not AI will impact almost every aspect of our lives in the next few years. The question is can AI be better than humans? We know that ChatGPT is around the 90th percentile of the law school admission test (LSAT) and medical school admission test (MCAT). There are also multiple examples where LLM’s significantly outperform humans. It is safe to say that AI companies such as Anthropic or OpenAI are developing tools that are replacing knowledge-based tasks at an alarming rate. This trend has taken off in the past few years and will only continue to accelerate in the future.
The SR&ED consulting field will be affected by AI in a material way. Customers that use AI tools to create a delightful customer experience will win. Sure you’ve been told by big accounting firms that Joe has 25 years of experience filing SR&ED claims and knows everything about a field of science. The fact of the matter is that Joe simply cannot keep up with the pace of developments, takes a vacation, is moved by his company, is unavailable when you want him to be and cannot hold a candle to a properly supervised AI system.
Supervising the AI is critical. It is wholly insufficient to open ChatGPT, send it all your most valuable R&D data (which it will retain and use to train its model if you don’t know what you’re doing), and expect it to create an SR&ED claim. It also does not make sense to build your own machine learning system when you anonymize client information, put it in a data lake, and set a machine learning algorithm to come up with a system for writing SR&ED claims. The whole premise of SR&ED is that the solution cannot be easily found in the public domain. Therefore, trying to train a model to understand patterns or uncertainties it has never seen before does not work. This is why no company has been successful in developing an ML system for SR&ED consulting. The secret sauce is in experts supervising the AI. Not only must humans effectively guide it, they must continually work to improve its performance.
Summary
Technology supporting SR&ED consulting companies has come a long way. The tools are starting to be a resource for clients rather than for the consultants. In 2025 we will see AI continue to impact all forms of consulting. AI already has a significant impact on doctors, lawyers and many other professionals. By utilizing supervised AI, consultants can offer better service at a lower cost with more flexibility. It is really a win-win for everyone as companies can retain more of their SR&ED claim to fuel further R&D. The best part about this evolution is that the quality of the SR&ED claim is better now than it has ever been. With AI understanding more about a specific company’s technology than any one individual consultant, it is poised to completely replace technical consultants. However, due to the complexity of the SR&ED program, humans will need to supervise the process and the AI to ensure that it is compliant with the CRA guidelines. Soon, compliance will also be done by AI faster and better than by humans.
The future of SR&ED consulting is here, where technology and AI are transforming the landscape for good. Ready to stay ahead and maximize your claim? Partner with experts who are harnessing the latest in supervised AI to simplify and optimize your SR&ED process. Contact us today to see how GrowWise Partners can support your innovation journey with ease and efficiency!
