People discussing ideas around the round table

How AI-powered data optimization drives tax and finance transformation

Related topics

AI and a standardized approach to data generation can help transform people and process across the wider tax and finance function.


In brief

  • Data is fundamental to the tax and finance function achieving compliance in an increasingly complex reporting environment.
  • Most businesses, still struggle with extracting high-quality data in an efficient and sustainable way.
  • AI data reconciliation as well as a centralized and automated approach to data can liberate tax teams from compliance work so they can add greater value.

Data is transforming tax and finance functions, helping teams switch focus from routine compliance work to become strategic, data-powered generators of insight, capable of guiding the wider organization. This transformation is particularly important considering the speed and scale of regulatory change and the need to share ever-greater volumes of granular data electronically with tax and finance authorities.

Accessing and operationalizing high-quality data at speed, however, continues to be a challenge for most organizations. The latest EY Tax and Finance Operations Survey reveals the extent of this challenge, with tax personnel currently spending three-quarters of their time (75%) on routine compliance work, which includes data collection and cleansing, tax return compliance and related reconciliations. Meanwhile, tax practitioners spend just 28% of their time on higher-value work such as data analysis, tax planning, managing tax controversy, general strategy, communications and risk management.

The key to unleashing the full transformative power of data lies in enabling tax teams to access the high-quality information they need. This is no simple task. Nearly half (48%) of organizations cite the lack of a sustainable plan for data and technology as the biggest barrier to achieving their vision for a modern tax and finance function.

Forward-thinking organizations are now using a range of strategies to overcome this data challenge, including centralizing and improving data creation at source and using artificial intelligence (AI) to pinpoint low-confidence data, so it can be escalated for people review.

Creating a single view of tax truth

Terri Beigh, EY Partner, Tax Technology and Transformation, Ernst & Young US LLP, is currently collaborating with Microsoft and one of the world’s largest manufacturing companies to help overcome this data challenge and transform the organization’s tax and finance function. She says the solution is to achieve a single view of “tax truth” which can then be used with confidence across multiple teams and processes.

Beigh explains that standard practice typically involves individual tax and finance teams retrieving the information they require to satisfy their compliance requirements.

This means that multiple teams requiring a trial balance, for example, often run completely different T-codes, resulting in different figures. Tax and finance teams must then engage in a potentially protracted retrospective reconciliation process. As a result, the process of data operationalization can become somewhat adversarial, according to Beigh.

“When locally generated data is shared, each team usually thinks their trial balance is the most accurate, while every other team’s data is wrong,” Beigh says. “Each team has their own tried-and-tested way of doing things. Teams can still achieve their compliance requirements, but this process lacks accuracy and it’s inefficient.”

This local generation of data also exposes differences in layout and language across jurisdictions. For example, every time a finance director wants to generate a trial balance, they must provide it in multiple formats and layouts, drawing out a complicated process even further.

Standardizing and centralizing tax and finance data

The solution for Beigh’s team, Microsoft and client-side collaborators, is to eradicate duplication by centralizing the data generation process and tracking core data points back to their original source.

“We embarked on a journey of data standardization and centralization,” Beigh says. “This involved establishing how many teams needed a trial balance, agreeing a standard format with the same parameters and the same periods of use.

“Instead of five versions of a trial balance, every team starts with the same one, pulled centrally at the same time and given to the teams at the same time.”

The result is a single data pipeline stretching all the way back to the general ledger entries for all detailed financial transactions, with Application Programming Interfaces (API) connectors automatically retrieving data from SAP using Microsoft Finance Insights solution.

This new data-generation process has superior controls and is more accurate due to a vast reduction in manual handling. It is also faster and less labor and resource intensive.

Mariusz Beben, Senior Director Microsoft, Industry Solutions Delivery, explains that automated data extraction played a key role.

“We effectively replaced hundreds of monthly manual information requests with scheduled and automated data pulls, backed up with automated quality checks. This gives tax teams more time to query and reconcile any data discrepancies before a tax return is filed, rather than after,” Beben says. 

We effectively replaced hundreds of monthly manual information requests with scheduled and automated data pulls, backed up with automated quality checks.

This, he explains, results in a process that is smoother, more streamlined, timelier and less-error prone. This approach, which relies on Microsoft Finance Insights, can also be easily scaled without the need for additional headcount or resource.

A centralized and standardized model also ensures that organizations can more easily maintain an overview of the information they share globally with tax authorities and regulators, ensuring their data contributes to a unified narrative.

Unleashing tax and finance transformation

By eliminating the time spent on data collection and manipulation, Beigh says her team and their collaborators have accelerated the tax filing process by 20%-30%.

Previously, preparation of tax returns continued right up until the filing deadline set by tax authorities. Beigh’s client now has significantly more time, enabling everyone to focus on broader tax matters and managing controversy, which leverages their tax knowledge in innovative ways. It also helps them keep up-to-date with rapid legislative and regulatory changes.

“With a solid foundation of unified data, it’s possible to analyze, plan and make strategic decisions with greater confidence because you’re using the same definition of net income across all entities,” Beigh says.

With a solid foundation of unified data, it’s possible to analyze, plan and make strategic decisions with greater confidence because you’re using the same definition of net income across all entities.

She says this centralized and standardized approach to data generation can trigger transformation at a foundational level.

 

“It’s about taking the organization to a higher level. We’re helping tax teams embrace new forms of technology so they can execute the work they’ve been asked to do today, but it’s also about upping their technology skill sets and giving them the data they need to act and think in a more strategic way.”

 

This organic use of data aims to pull the tax and finance functions together, transforming people and processes as technology is enhanced. This transformational approach will prove critical if tax teams are to meet tax authorities’ demands, switching to real-time data exchanges instead of the existing tax return process.

 

Using AI to “find the needle in the haystack”

 

Creating a single view of the truth by centralizing and standardizing data at its source is a powerful and future-proofed way to address the data challenge. This process also includes tax functions harnessing AI solutions to identify low-confidence data.

 

Ivan Roussev, an EY senior manager in the Tax Technology and Transformation practice at Ernst & Young US LLP, says: “Tax work is a ultimately a bit like looking for a needle in a haystack. We’re constantly looking for errors hidden in a mass of information. But what if we could quickly and easily wave through the majority of high-confidence data and focus our time on the inaccurate information? That’s what we’re doing with artificial intelligence.”

 

Roussev explains that the bulk of tax accounting reconciliation work involves a small number of transactions. These transactions are relatively simple to reconcile because they are easily identified, easy to understand and can be processed automatically using encoded deterministic rules.

We’re constantly looking for errors hidden in a mass of information.

There is also a subset of data points which tend to be more problematic, however, the proverbial needle in a haystack. AI can be particularly good at identifying this subset, however, thanks to its ability to learn from patterns hidden within historic data. Having identified potential data anomalies AI can then generate a percentage confidence score for each data point. High-confidence transactions within this subset go ahead unchallenged, while low-confidence transactions are automatically escalated for people review.

Roussev explains that clients using this two-step AI-powered reconciliation process have successfully reduced time spent on reconciliation from thousands of hours annually to just tens of hours. This approach also enables monthly rather than annual tax accounting reconciliations. As a result, accounting errors are discovered earlier, they trigger fewer secondary issues, and findings are quickly fed back into AI models so they can learn, making them even more effective in future.

AI can also be trained to categorize high volumes of VAT transactions. According to Roussev, the majority of tax errors can be traced back to low-quality purchase order data (namely commodity codes and material groups) entered by employees who do not necessarily appreciate how important this information is to determining correct tax treatment. As a result companies routinely pay either too much or too little tax on high volumes of transactions.

Historically companies have addressed this issue by employing auditing firms to conduct reverse audits (often on a three-year cycle) to carefully review their tax payments and determine if they have overpaid.

Now, however, AI is increasingly being used to automatically identify and correct faulty tax categorization. This not only dramatically increases time to value, it also significantly boosts confidence in tax data, enabling companies to make better business decisions faster and ensuring funds stay within the organization and don’t need to be held in reserve for tax adjustments. Furthermore, AI can also help identify under-payments, which are often outside the scope of a reverse audit.  We can think of AI as a junior member of the tax team, streamlining processes and providing valuable insights to human team members. The technology is also poised to be a liberating force for tax practitioners giving them additional bandwidth for innovation and other value-added activity.

Improving the quality of data in real time

The next step, according to Beigh and Roussev, is to harness artificial intelligence to improve the quality of tax and finance data in real time and at source, with the aim of further reducing the need for reconciliation and recategorization exercises.

Once achieved, this would allow the tax function even more time to engage with the wider business as it continues to evolve and change alongside markets, products and services. With access to higher-quality data the tax function will be able to proactively influence the business at a much earlier stage, rather than being consulted at various points along the journey.

Summary

Data is the foundation on which the transformation of the tax and finance function is currently being built. However, the current practice of local teams extracting and generating their own tax data – which is slow, inefficient and inaccurate – is a major speed bump in this journey. The EY organization has collaborated with Microsoft and some of the world’s largest organizations to automate and improve tax data generation and extraction.  

Related articles

    About this article