FP&A meets artificial intelligenceGet ready for the fundamental changes AI may bring to financial planning and analysis.
One purpose of artificial intelligence (AI) is to capture, manage, and analyse amounts of data too large for less advanced technology to handle, and to produce insights faster and more accurately than people can. While AI promises to help businesses prepare better for potential opportunities and risks, the technology presents a challenge for management accountants in financial planning and analysis (FP&A).
FP&A teams provide the data analyses that top management asks for to make business decisions. With the potential for AI to handle varying degrees of that data analysis, the role of FP&A could change fundamentally and require management accountants in FP&A to broaden their skills.
Decisions based on a large number of measurements will be repeatable and scalable, and dynamic metrics will ensure everyone is on course, said Chris E. Ortega, a senior finance manager who is responsible for FP&A, finance, and accounting at Emarsys, a Vienna-based multinational company that provides marketing technology.
"Think of it as the CFO and FP&A leading a boat through waters with marketing, sales, operations, and client success in the boat as well," Ortega said.
AI targets FP&A at new opportunities for the business through the increasing number, quality, and accuracy of its insights.
"Perhaps AI will reduce the number of people needed for FP&A, but I think you will need professional oversight to focus information, provide management direction, and incorporate the planning and management decision-making process," said Mick Armstrong, CPA, CGMA, the CFO of Micro 100, a toolmaker based in the western US state of Idaho.
To provide the correct level of business intelligence or decision support, an FP&A team member would benefit from developing his or her critical-thinking/analytical capabilities and learning to work collaboratively across finance and nonfinance teams, said Nadim Ahmad, former CFO at Data Communications Company, a UK-based company implementing a secure data and national telecommunications network.
"One route to empowering the FP&A function is for the CFO to ensure the relevant communication and soft skills training is given to the team," Ahmad said.
Challenges in using AI for FP&A
Getting CFO buy-in for AI for FP&A can be challenging.
"It may not be simple to push through a business case for AI if there are other priorities. The strategic value of AI may be hard to quantify tangibly," Ahmad said. "It would be a tough choice for a CFO to make, though that could change as a CFO's insights into technology and, more specifically, AI improve over time."
"The ability to pull data faster and begin the analysis period sooner in the close cycle will be vastly important as everything in our world speeds up," said Marie Masenga, CPA, CGMA, global corporate controller, Market Data Services Ltd., an information technology and services company in Phoenix.
But there are no guarantees that you will get an AI process or algorithm that does a better job on a problem than people are already doing. "Whether you get a process or an algorithm that does a good job depends somewhat on the type of business, the sector, and the kinds of decisions you are making. If you are in financial services or insurance, where your business is very computational, very calculational, then building good algorithms should be easier," Ahmad said. "In large telecoms, for example, getting an algorithm to compute a strategic decision would be more challenging, particularly where external factors such as consumer or competitor behaviours need to be considered before finalising an outcome. You would need a hybrid approach where AI takes you to a certain point in your calculations, and your people stress-test and evaluate that data and make the final decisions."
Data quality is an essential challenge to using AI for FP&A. "AI, like all technology, is only as good as the information we feed it. While reporting and forecasting can speed up, the data quality entering the system still needs to be a strong focus for accounting teams in every organisation," Masenga said.
If the benefits of AI do not outshine the risks for your organisation, wait a few years. The technology is evolving with every passing year. Algorithms that do not make a significant difference in your sector now are likely to improve over time. And one day you may be using AI and not even know it. According to Gartner, AI will be in almost every new software product by 2020.
Here are some best practices for using AI for FP&A:
- Apply AI/data scientists to FP&A business opportunities and challenges: This will help you achieve more immediate improvements. "The best change comes from addressing real business problems. It's best to have a real problem and arrive at a solution to see whether AI will be an enabler for solving problems more efficiently and effectively," Ahmad said.
- Do not let data quality limits hold you back: Rather than put off AI, start with small data sets that meet quality standards. "CFOs should not wait until their data is perfect, but they need to make sure they are stress-testing the data and running parallel processes so they meet a minimum standard for data integrity before they move to an AI solution," Ahmad said.
- Pair your data scientists with the FP&A team: Use AI with a group that consists of data scientists and FP&A professionals working near one another. "I've seen businesses where the data scientists report to the operations director or the sales director. That is not a sensible move. If the data scientists' purpose is to generate enterprise-wide data for strategic or commercial decision-making, they should sit with the FP&A team or, at the very least, with a team under the accountability of the CFO," Ahmad said.
- Ensure that the data scientists know your business strategy and objectives: "If the data science teams are providing information to FP&A that is useful for decision-making, then they need to understand the corporate vision and strategy, in order to identify the optimal data sets required for making the right decisions," Ahmad said.
David Geer is a freelance writer based in the US. To comment on this article or to suggest an idea for another article, contact Sabine Vollmer, an FM magazine senior editor, at Sabine.Vollmer@aicpa-cima.com.