FEI Daily: Can you please describe how CognitiveTax Insight (“CogTax”) works and provide some use cases?
Andy Gold: CogTax leverages machine learning and smart OCR to ingest both unstructured data (e.g. data points from invoices, from POs, and non-ERP, non-systems data) and structured data from a company’s purchasing systems and tax engines. With the aid of a human who trains the cognitive technology, it applies machine learning technology to that robust amount of data.
In the training, an individual will interact with the cognitive technology and “teach” the machine learning taxability determinations based upon the client’s facts and transactional information available. Over time, the system learns as it sees more and more transactions that are substantially similar with the goal of providing a highly accurate solution in terms of being able to identify, in real-time, whether a transaction is taxable.
This technology is highly beneficial in the area of sales and use tax where companies historically will have thousands or even millions of transactions in a year where some tax determination has been made where they either paid or accrued tax on the purchase of goods. Given the volume, they can’t always get those tax determinations right. Currently, a lot of companies do a sampling across their transactions to manually determine if there has been an over or underpayment of tax, and then make some estimates across the entire population of transactions to try to recover overpaid taxes.
We feel that leveraging cognitive in a space like sales and use tax, where there’s a significant volume of data and where companies have historically had to be backward looking, places the user in a more proactive position – where they can get in front of these tax determinations on a more current or forward-looking basis with a higher degree of confidence.
FEI Daily: Beyond sales and use tax, are there other ways that you’re hoping to leverage this technology?
Gold: We think that cognitive has broad applications in many areas of our profession, especially those dealing with high volumes of data and determinations that are based on certain information but ultimately come down to a classification decision (i.e. is it taxable, is it exempt, does this qualify as a tax credit, is this a qualifying R&D expense or not). We see applicability in the broader indirect tax space (beyond sales and use tax), VAT, and other foreign transaction taxes (e.g. Canadian GST). Also, as you move into other parts of tax where you have classification needs, you see this in a variety of income tax calculations and in state tax calculations.
It’s important to note that this technology doesn’t remove the human from the process. There is still a high degree of subjectivity in tax laws and their application. A lot these types of determinations are more complicated than these simple classification-type decisions. This technology is a tool to help the user focus in on decisions and bring a level of judgment.
FEI Daily: What other technologies is Deloitte working on as part of its ‘Tax in 2020’ initiative?
Gold: We’re pretty excited about our 2020 initiative, and we view it as the digitization of tax. We’re thinking about all the things that we do and our clients’ user experience working with us and we’re migrating to a more seamless experience from the very beginning with data collection through to the filing of a tax return or preparation of other tax calculations. Imagine concise data collection from a client’s data sources (e.g. ERP systems, subledgers) combined with data wrangling tools to transform the data into the proper usable format in a way that makes it more complete and rich. Then we layer on analytics, to help surface insights and anomalies. Ultimately feeding that information into reporting tools and tax engines which generate reports used for various tax purposes (e.g. compliance or planning).
This digitization is occurring for all areas of tax: indirect, federal, and state and local income taxes, and we feel that it is important to invest in a broad technology platform that enables all of those capabilities to streamline processes to improve the ability to get at the information that matters – thereby moving our clients into the position of being able to focus on analytics and apply professional judgment and experience to all of this data. As opposed to spending so much of their time on dealing with and manipulating data. The initiative focuses on bringing value, solving problems, and surfacing insights.
FEI Daily: How will the digitization of tax impact the skillset required for future tax professionals?
Gold: For years, we’ve been talking to our professionals about wanting them to continue to advance their consulting and analytical capabilities – rather than just focusing on getting the tax compliance done. The digitization of tax is really going to accelerate that process; because, it’s going to move them out of the data process like work and out of some of the more time consuming, repetitive tasks by providing these analytical tools that empower them to see trends, run scenarios, and be more analytical. Moving forward, tax professionals will need to have the following skills:
- Apply data extraction tools;
- Leverage data wrangling tools; and
- Use visualization tools and dashboards
Really, the tax professional of the future is going to need all of these skills while maintaining and placing an imperative on subject matter expertise in technical tax subjects. To truly add value, it comes down to expertise and to a professional’s ability to innovate, create value, and solve problems.
FEI Daily: Looking ahead five years from now, how will the tools in tax departments differ compared to those today?
Gold: The emerging tools that we’ve discussed in this interview are going to be very mainstream (i.e. tools for data wrangling, dashboards, and analytical tools). It will become very typical to see companies leveraging automation technology – whether you think of that as RPA or data wrangling tools.
When I poll audiences in speaking engagements, a third or half are familiar with the types of technologies we’ve discussed. Two years ago, it was probably 15%. Over the next three years, it’ll be everybody in the room.
Then I think that you’re going to see more integration of the technologies, and I think that you’re going to see more expansion in areas like cognitive technologies – where it’s not just the repetition of manual tasks that the tools are performing, but some level of predictive analysis and the ability to help guide a decision, like our cognitive technology.
FEI Daily: In the digitization of tax, what are the biggest challenges that tax departments can anticipate?
Gold: In working with clients today, I think they struggle with what’s possible. They say, “I hear about these technologies. I’m interested. I’m even committed to advancing the use of these emerging technologies in my department, but I don’t know what’s available.” They need to learn more about the art of the possible. So, first they need to understand the various available technologies – then they have to understand how these technologies are being applied to bring value to organizations.
Companies struggle with knowing how to get started. Where do you start? You don’t boil the ocean, you need to pick an area that people will care about and where you can have success.