The pace at which finance leaders must change course and re-forecast is increasing dramatically. Global events, competitive tactics, marketplace dynamics, and geopolitical moves create an influx of information and drivers that impact organizational tactics and decisions. Given so much uncertainty, it’s not possible to plan for just one or two future scenarios. Smart companies make sure they can manage several financial scenarios in a framework where they can make decisions with agility and confidence.
To do that, finance leaders are turning to artificial intelligence (AI) to uncover richer intelligence that will help them develop more accurate forecasts. From there, they can make more frequent, faster decisions with greater determination.
Of course, finance isn’t the only corporate area where AI has brought new insights and disruption. But compared to other functions such as sales and the supply chain, finance poses unique challenges for adopting AI. Finance is highly regulated, so its operations need to be accurate down to the last decimal place, and every step explainable. Those expectations can sometimes clash with “black box” AI operations. Also, there’s a high premium on predictability. But AI’s complex algorithms can sometimes create dramatically different outputs from relatively minor changes to certain inputs (the so-called “butterfly effect”).
So adapting AI effectively to finance requires forethought about how and why you’re using it. To help you make your finance organization future-ready, here are five best practices for adopting AI to better manage volatility:
1) Use AI to navigate uncertainty
My colleague and Anaplan’s Chief Data Scientist, Tal Segalov, explains that AI actually thrives in uncertainty. If things were always clear, simple, and predictable, humans wouldn’t need much extra computing power to make decisions; when they’re not, that’s when AI shines. It gives planners an edge in four key ways:
- Agile and continuous: AI is able to react to new data – from both internal and external sources – and generate new predictions in a fraction of the time that manual analysis would have taken.
- Scale: AI can scale to increasingly high loads of data across line items, time series, cues, business units, and more – in greater levels of granularity.
- Data: AI unlocks the power of data signals from thousands of different attributes and data streams to reach a decision.
- Deeper time horizon: AI helps extend the time horizon of your forecast to plan further into the future.
2) The long-range plan (as we know it) is obsolete
One great use case for AI’s powers is forecasting. And it’s arrived just in time to help drive the traditional annual planning cycle into extinction. Finance leaders know annual planning is a dinosaur – time-consuming, resource-hungry, and too slow and rigid to keep pace with the rate of change. And it doesn’t provide the level of accurate insight planners need. Adopt long-range rolling forecasts infused with predictive intelligence to help you make decisions faster and with confidence.
I recently spoke with Nicolina Saporito, Managing Partner at EY, about this required shift. She sees that many companies are already adapting in two important ways:
- They are increasing the frequency of their strategic planning capabilities. Importantly, they are augmenting, not replacing, forecast capabilities.
- They are significantly increasing the number of scenarios that evaluate various sensitivities across different assumptions. With that information, they can develop a catalog of risks and opportunities that facilitate faster decision-making in a way that’s less like a single plan and more like a series of playbooks.
3) Expand your data sets to better account for volatility
It’s important to understand the key drivers for achieving your operational plan, know which of those drivers are going to be most volatile and uncertain, and make sure you have your fingers on the pulse of those drivers.
As Nicolina explained:
“Forecasts are traditionally managed around organizational or account structure, such as a business unit or geography. But more recently, to respond to so much market disruption and volatility, firms are expanding their data sets to include more granularity and new dimensions—for example, product, customer, and channel dimensions that are much more vulnerable and sensitive to those types of disruptions. That is really critical to understanding the impact and how that then translates to your finance and operational metrics.”
AI can help with building in those new data dimensions and levels of detail.
Tal offered this advice:
“There’s no such thing as too much data (for AI-based predictive analysis). Bring it on! On the other hand, too little data is not a huge worry. Modern systems, especially when connected to Anaplan, can leverage huge amounts of historical data that spans multiple angles. You can combine all of those different viewpoints into one coherent forecast that spans all of those viewpoints. As a result, (that combined) forecast tends to be much more accurate than each of the single underlying elements that made it.”
4) Shift your perspective on transformation
With the transformative effects of AI, it’s easy to focus on data to improve planning and forget about the organizational changes involved.
Nicolina put it this way:
“Don’t underestimate change management. We’re not just changing the plan; we’re changing the whole process and the way we interact with information. The skills needed for that are quite different than those needed to manage 1,000 spreadsheets. That’s part of the journey, but it’s also a tremendous opportunity to move to advanced digital platforms. Most people I know in finance don’t enjoy aggregating 1,000 spreadsheets! Clients are excited about this because it’s changing the job in a way that’s beneficial to them both personally and professionally.”
5) Use a bottoms-up AND top-down approach to intelligent forecasting
To ensure the accuracy of predictive forecasting efforts, it’s important that your AI-driven forecasting works from the bottom up and out to the edges in your organization, in parallel with a top-down focus on a few key things that are driving the most impact. In other words, you need to look at both very granular, detailed data from various sources such as ERP updates on manufacturing output or customer profiles from a CRM system (the “bottom-up”) part, as well as high-level historical data for the functions you’re trying to predict (the “top-down” part).
For example, if your sales forecast doesn’t look positive, that could be due to the quality of current pipeline opportunities (traceable from details in your CRM opportunities), or from long-term seasonal fluctuation (based on analyzing historical patterns). Your actions in response would be very different based on the data sources, even if the goal is always to improve the forecast. (In this case, if the AI says the problem is pipeline opportunity quality, you’d spend more on top-of-funnel lead generation; if it’s seasonality, you might adjust sales incentives.)
With this multi-directional approach, the capability to leverage and amplify data science and AI is no longer left to one specialized department or team. It’s democratized across all groups. Companies that can do this fluidly are going to have a competitive edge.
AI is evolving, rapidly. It’s made significant leaps in sophistication and in aligning more with the needs of business users. Yes, the future ahead can seem a bit intimidating considering the breadth of uncertainty. But as Nicolina said, “It’s a journey, a progression — not a zero-to-60 jump.”
The opportunity energizes me, and I know that AI will play an essential role in giving us more confidence.
If you’re curious, learn more about our AI-driven Predictive Insights solution.