Month: Dec 2016

Closing the Gap: Predictive Analytics In Action

By Nilly Essaides

Many finance functions are still struggling with the adoption of advanced analytical tools. Several surveys show the adoption of predictive analytics remains low, typically under 20%. However, that’s going to change as:

  • FP&A is increasingly called upon to deliver data-driven decision support to its business partners, i.e., not just explain the “why” things happen but “what’s going to happen next.”
  • Big data becomes more available and accessible.
  • There’s a proliferation of available predictive analytics tools.

Becoming Table Stakes

This ability to look into the future separates top performers from peers in the EPM space. Already, there’s a leveling of the playing field. As the chart below shows (based on a 2014 performance study by The Hackett Group but it holds true today) in some areas (look to the left) peers have quickly caught up to top performers, for example in areas like scorecards, predictive modeling and self-service analytics and reporting.

But look to the right and you’ll see some big gaps. Multidimensional analysis and risk analysis are very good examples where top performers are ahead.

Case in Point: Risk Analysis

It all sounds very technical and complicated, but it’s important to remember that areas such as risk analysis come down to very straightforward applications that companies already use today to solve real-life business problems.

Here are a couple of examples.

At one company, predictive analytics are being applied in the credit and collection space to help finance make the collection process more efficient by analyzing the risk (or lack thereof) of customers paying their bills on time.

Ordinarily, based on what’s customary terms for the market, this company sets out a policy whereby it notifies customers electronically or through the mail before their bills are due. So, for example, if the terms are 55 days, this company will send a reminder at day 50. Having recently implemented an Oracle solution domestically (which has allowed it to collect a richer level of data about when and how customers pay), it has experimented with a predictive analytics solution that allowed it to assess when customers are likely to pay.

So, instead of spending money and time on sending an alert on day 50,

“If we know they’re going to pay on day 54, we can avoid sending the notice,” said a finance executive at the organization.

By crunching historical data and finding patterns, this finance team has been able to predict the payment behaviors of its customers reducing the cost of the collection process. “This is still in the process of evolution,” he said. Eventually, they plan to roll it out throughout the global enterprise.

For now, they’ve not settled on a predictive analytics tool. “We’re looking at multiple option,” he said. “We’re purely in the exploration mode.” Ideally, this team wants a predictive tool that would “write” directly into its ERP, instead of produce a report that would then leave it up to them to figure out what actions need to be taken. The result is likely to be the use of multiple tools for capturing, managing and analyzing the data in a heterogeneous environment.

The leader of the FP&A team at another company has put predictive analytics tools to use in helping a business unit assess the default risk in its consumer loan portfolio. By looking at past default instances, the tool analyzes and predicts the subsequent likelihood of default, allowing the business to make savvy decisions about how much reserves to put aside for coming periods.

“What drove the initiative is that we wanted FP&A teams to work more collaboratively with business partners and offer insight to the business,” said this FP&A executive. “We wanted to have a model that is based on data and science.”

The lessons for EPM executives are as follows:

  1. While adoption rate today may be relatively low, this is bound to change. To become a top performer, start with adopting predictive tools in the more common areas and then graduate to less explored areas like multidimensional analytics and risk analytics.
  2. To roll-out a successful project begin with a pilot. In both companies’ case, FP&A started small, focusing on one process in one locality and was tool agnostic. It was more important to experiment than to make big decisions about technology.
  3. Finally, work closely with the business user. Predictive analytics are only as valuable as the insight they deliver to the business. The result needs to be actionable information that adds value to the organization. In the case of the collection process, the new capabilities saved real dollars. In the case of the loan default rate, FP&A was able to provide the business with the information it needed to plan better.

Learn More

Source: Generation CFO LI Group