As a finance executive, do you always trust the forecast from a knowledgeable FP&A staffer over one given by a sophisticated, machine-learning-enabled forecasting tool? Are you satisfied when a number is close enough for business unit decision-making purposes, such as assessing a market trend, or does all data need to be pinpoint accurate before being released to operating units?
If you answered “yes” to the first question and “no” to the second, you may have to let go of two outmoded notions — one about technology and the other about data, according to analysts at IT research firm Gartner. You’re also not alone. To succeed as digital transformation leaders, finance chiefs will have to adjust their mindsets, at least in some contexts.
“Without the right mindset from leadership, it won’t matter how much talent or budget CFOs throw at digital transformation,” said Dennis Gannon, a Gartner vice president, research.
Speaking at the Gartner CFO & Finance Executive conference in National Harbor, Md., this week, Gannon said the first necessary change is trusting technology more as finance moves from automation to “autonomous finance” — a finance function “partly governed, and majority operated, by self-learning software agents that optimize front-, middle- and back-office operations.”
For example, Gannon told the story of a company that instituted forecasting software that incorporated macroeconomic, market, and enterprise transactional data to backstop its planning process. “The new process is amazing, works well, and does much better than humans ever did,” Gannon said. But finance isn’t using it yet, because they don’t trust it.
“CFOs tend to see technology as a tool but rely on people to make decisions,” said Gannon. “Even when the evidence shows that technology makes better or more accurate decisions, people are still reluctant to trust it.”
The technical term is “algorithm aversion” — a phenomenon “where humans reject advice from an algorithm in a case where they would accept the same advice if they thought it was coming from a person,” according to Wikipedia.
Even when the evidence shows that technology makes better or more accurate decisions, people are still reluctant to trust it. — Dennis Gannon
The onus for change is not entirely on the CFO, however.
“As finance leaders, we need to know if the work being done by an algorithm can stand up to an audit, can survive regulatory scrutiny, and [won’t] introduce new and unknown risks into our organization,” said Gannon. So, it's not enough for an algorithm to simply be correct and accurate. “It needs to be more than that, to earn our trust,” Gannon said.
Besides being accurate, emerging technologies need to be fair, explainable, understandable, accountable, secure, and safe to earn the trust of finance executives.
Data Governance Shift
The second mindset change concerns the “command and control” governance of data across the enterprise. Finance’s “accuracy-first” mindset toward data governance is critical when reporting financial data to the board of directors, regulators, investors, and other stakeholders.
“A good day’s work for finance is when its work is error-free,” said Nick Duffy, director, advisory, for financial and data analytics at Gartner. For example, according to Gartner's research, the data a finance team produces during the close is typically more than 99% accurate.
“We have poured resources into source systems, ensuring [data] governance is tightly controlled,” Duffy said. “We’ve laid the foundation for a single version of the truth” — when the data in source systems, the data repository, and reports and business intelligence tools are in agreement.
But while finance is focused on accuracy, and sometimes “accuracy at all costs,” business partners aren’t satisfied with the data that finance generates for operational decision-making, Duffy said.
In a Gartner survey, only 13% of business partners thought the nonfinancial data from finance truly reflected business performance, only 21% felt the data was complete, and 26% thought the data was up to date.
Obsessing over data accuracy in operational decision-making can mean sacrificing transparency, completeness, and timeliness, Duffy said, as operations demand more data and information from finance to make decisions in a complex, interconnected world.
Finance may build a highly accurate report, and decision-makers can have confidence in every data point, but the data may arrive “way too late” for timely decision-making, Duffy said.
Sufficient Data Governance
To adapt, CFOs have to think about adopting “sufficient data governance,” as opposed to a single version of the truth, said Duffy.
Sufficient data governance is a shift away from accuracy as the sole, overriding objective. It uses different levels of governance for different data types and uses. Reports and business intelligence tools, for example, deliver accurate and consistent data where needed, and fast and relevant data everywhere else.
For the data that executive management and the board used to drive enterprise and strategic outcomes, a centralized, universal approach to data governance is still required. But for tactical outcomes and operational decision-making — like the information operations needs to diagnose performance issues, monitor market shifts, or spot business risks — the objective of data governance should be efficiency, Duffy said.
The problem is that only about a quarter of finance departments use the “sufficient” approach, Duffy said. Data governance needs to shift from being just focused on accuracy to data being “optimized and ready,” Duffy said.
