Today, data analytics solutions offer benefits to practically every field of expertise imaginable. If deployed wisely, data analytics can play a large role in obtaining insights into large amounts of data. Fortunately, in the field of taxation, data happens to be in abundance, be it structured data stored in ERP systems or unstructured in the form of court decisions and contracts. This blog aims to set some expectations for students of Tax & Technology with regards to  the use of data analytics within their upcoming profession.

But first, what do we actually mean by the term ‘data analytics? Artificial intelligence (A.I.)? Excel on steroids? Data visualization through dashboards? Simply put, data analytics is the scientific process of discovering and communicating (meaningful) patterns which can be found in data. The process often covers (a combination of) retrieving, inspecting, cleansing, transforming, modeling and visualization of data.

"Fortunately, in the field of taxation, data happens to be in abundance"

The goal of this process is to automatically provide answers to questions that lay hidden within data. Depending on the business case and the required level of sophistication of the data analytics solution, questions may hold values ranging from descriptive (“What happened?”), diagnostic (“Why did it happen?”), predictive (“What will happen?”) to prescriptive (“What should I do?”).

A quick heads up for those A.I. enthusiasts amongst you, please note that analytics may very well include aspects of A.I., however, it is often thrown into the mix wrongfully. Therefore, please do not be disappointed if you find yourself ploughing through MS Excel extracts after finally landing that Tax Data Science job at Company X advertising with state-of-the-art tax data analytics solutions.

So, what can you expect as a tax professional working on a data analytics solution? These days, within tax, your typical data analytics solutions are mostly build upon structured data and focused on descriptive, diagnostic and predictive questions like “What is my current VAT exposure?”, “Why did my taxable base in country X increase?” or “What will be our likely tax exposure in year X?”. These types of data analytics solutions are relatively common and provide companies with insights in their tax positions through so-called ‘dashboards’ contributing to their Tax Control Framework.

More exciting, in my opinion, are the recent advantages in applications of A.I. such as Natural Language Processing. These technologies potentially allow us to automatically answer questions in relation to tax matters that lay hidden within unstructured data. Questions like: “Which cases should I consider in my research given situation X?”, “What are the questionable provisions in this M&A agreement?” , “What will happen if we structure our company or transaction given fact Y?” or “Should I appeal to the judgment of the Court in case Z given development Y?”. Automation to this extent could, for example, significantly increase quality and efficiency of legal research and impact our daily labor as tax professionals. Is this the future?

Collaborations between man and machine to tackle cognitive challenges have been around for a while. Ever since IBM’s Deep Blue defeated former world champion Kasparov in a chess game in 1996, there have been suggestions that A.I. is going to replace human labor in the near future. But Kasparov saw things differently: instead he foresaw a future with collaboration between A.I. and humans. He started up Centaur, a new form of chess in which human players use A.I. to predict the outcomes on the game board. It turned out that this collaboration outperformed both sole efforts by A.I. and humans.

"Collaborations between man and machine to tackle cognitive challenges have been around for a while."

The inflation of human efforts in chess continues as we progress in time and enter the deep learning era. Nowadays, Google DeepMind’s Alpha Zero effortlessly beats teams of human/A.I. players. Due to the relatively straightforward and limited variables in the reality of a chess game (e.g. set rules and reward system - winning is good/ losing is bad), deep learning algorithms can simulate and iterate over literally all the options imaginable and calculate which next move would statistically have the best chance of resulting into a win.

Unlike chess however, delivery of answers to tax queries is infinitely more complex than a game of chess as there are real world variables to consider. Additionally, when seeking the answer to a tax query, the answer sought after is often subjective and therefore not always clear and consistent. Not knowing what result to pursue makes it very hard to train deep learning algorithms. Moreover, approaching tax queries usually require a logical line of argumentation for the recipient to accept the answer. Fortunately (for us), the algorithms utilized for these predictions are often so-called ‘black boxes’ and do not (yet) provide any argumentation understandable for humans.

"Approaching tax queries usually require a logical line of argumentation for the recipient to accept the answer."

It is therefore imperative that results derived from such solutions are interpreted and checked by a tax professional. Fast forward a few years and, being a tax professional, could very well mean that your day to day tasks include auditing complicated tax prediction algorithms, or taking up arms against the arguments of a robot lawyer. But for now, it looks like we are entering an era where A.I. can function as an assisting tool allowing more efficient and higher quality tax advice.

In conclusion, my expectations would be that data analytics solutions continue to become primary tools for tax professionals of the future, allowing them to form unbiased opinions on tax questions, helping them understand key legal concepts better and see connections between like never before. Data analytics solutions can provide excellent support in this respect supporting tax professionals in finding relevant arguments, in substantiating advice and with statistical analyses. However, there are certain risks involved. It is therefore imperative that, today, both tax students and tax professionals start learning about these technologies and the risks and opportunities they bring forth.

‍