Practical Predictive Modeling to Improve Your Forecasts

“Next year, we need to do five percent better.”

“In the upcoming year, we should cut expenses by ten percent.”

As competition rises in the business world, companies continually need to boost revenues, reduce expenses, and become increasingly more profitable. But, if you’ve ever been in a budgeting meeting where statements like those above have been defined as goals, it’s likely you’ve been in situations where end-of-year results weren’t as good as leadership had hoped.

Blanket-statement goals can, in fact, be downright counterproductive. As just one example, cutting expenses by a certain percent could actually cut revenues even more, ultimately leading to lower profitability.

And, here’s another statement you’ve likely heard in company meetings: “There’s got to be a better way!” Fortunately, when it comes to creating forecasts that truly drive success, there is a better way: predictive modeling.

To illustrate how this could work, let’s use an international technology company as our example, one with tens of thousands of call center reps who service customers. The customer experience leader who oversees these operations needs to decide how to staff the call center, including how many people to have at what levels of knowledge and experience at what times of the day and week.

The leader will also need to schedule differently, say, in January when significant numbers of people recently received products as holiday gifts or purchased them to upgrade technology for the new year. She will also need to schedule differently when new products are launched and otherwise manage staffing based upon seasonal ebbs and flows.

Traditional Forecasting Models: Historical Data

Finance teams often use historical data to put together budgets and set goals for the following year. Using our call center example, they might automatically build in a certain percentage of funds for extra call staff each January and for the new product launch in July. In other words, past volume needs will serve as a driver for the company’s upcoming financial forecast.

The formula used might look something like this:

  • Add up the number of service calls received during each week of the previous year.
  • Determine how many were answered by the front-line team and how many needed to be passed on to someone else with more expertise.
  • Calculate the wages of everyone involved in each week’s servicing at the call center and use that to budget for the following year.

If the call center begins receiving significantly heavier or lighter volumes of calls when compared to historical data, this current market demand data can also be used to estimate volume need. In that case, the company could adjust staffing accordingly.

Then, at the end of each year, finance departments compare projections against actual expenses and revenues to determine how closely budgeting and forecasting goals were met before making next year’s projections. This forecasting model has been a longstanding methodology, so much so that this could be considered the standard approach for forecasting.

Disconnects in Language Used

Information gleaned from the standard forecasting model is typically discussed and relayed through language that accountants use, with references made to income statements, balance sheets, cash flow projections, and so forth. This makes perfect sense to finance professionals, but the information is often shared with people in lines of business such as sales and marketing, customer service, and information technology—and this is seldom the language they use in their own daily work.

Going back to our call center example, the customer experience leader overseeing this line of business is more likely monitoring the number of calls that come in during a shift, a day, or a week, among other quantitative measurements; quantitative and qualitative call center metrics may include these, for example:

  • first call resolution (FCR)
  • average handle time (AHT)
  • service level agreement (SLA)
  • customer satisfaction (CSAT) score

So, when you’ve got the finance team using accounting lingo and the call center using their own specialized lingo, a real disconnect can happen, making it harder to communicate and meet financial goals. Then, the sales and marketing team is probably talking about leads generated, quality lead percentages, and close rates, while the other lines of businesses are using their own niche metrics and jargon. The disconnect grows.

This brings up a key question. For forecasting purposes, how can a company effectively attach a cost to each of the quantitative measures in a way that makes sense to the teams involved? The call center leader, for example, needs to share exactly what this particular line of business needs to effectively play its part in meeting the organization’s needs overall. How can collaboration be created?

Intermediate Forecasting Approach

We’ve been discussing the standard forecasting approach that typically uses software right out of the box. Then there is an intermediate approach that harnesses predictive planning capabilities, giving the enterprise more statistical opportunities, with the process still driven by finance (as it is with the standard approach). Data inputs in the intermediate approach are similar to those used in the standard approach, typically information pulled from the general ledger or the company’s ERP solution. This can provide more guidance, but there’s an even more advanced approach to consider.

Advanced Forecasting Approach

This is where all can get especially interesting, because now the predictive planning functions are not necessarily owned by the finance department. Instead, they may be owned by the different business areas, such as IT, sales and marketing, as they provide their own non-traditional sources of data, which become inputs into the financial planning process.

So, in this model, the data sets used can be, but aren’t necessarily, financial. They can include customer sentiment data, metrics provided by sales and marketing, distribution data (on-time shipping, accuracy in order fulfillment, and warehouse capacity), and more. When using advanced forecasting models, a company must understand the non-traditional sources of data available, from both inside and outside of the organization, and then choose what’s needed to make the best forecasting decisions.

After identifying the best data sets, how should they be evaluated? What analytics-based decisions need to be made? These are the choices that drive a company’s unique modeling capabilities, and what’s important is to ensure that analysis is not being done for analyses’ sake, but to build bridges between lines of business, between finance and the lines of business, and so forth.

Opportunities with Common Language

Traditionally, the finance team sets targets across the entire company, typically with goals mandated by leadership, and then those numbers are pushed down through finance to other lines of business.

The sales team, for example, may receive their goals—and their response might be, “How can you expect me to sell more with fewer resources?” Discord, naturally enough, can set in, but if sales pushes back on finance, the response will likely be that set targets must be met to ensure company profitability.

Through advanced forecasting methods, though, predictive analytics and machine learning can be used to help the finance team manage risk and focus on profitability, while also turning these goals into practical and actionable steps that can be communicated to business lines in a way they can relate to. In other words, predictive dreams can be turned into profitable realities!

Advanced forecasting allows companies and their finance teams to move beyond what happened yesterday, past trying to figure out why yesterday happened, instead facilitating their way into how to have a better tomorrow. Predictive analytics allow enterprises to choose the most effective corrective actions needed to bring them back on track.

Using advanced statistical methods, business needs can be incorporated into financial models, providing inputs that allow for advanced analytics, optimized decision making, and a streamlined pathway to enterprise-level success.

Oracle EPM Solutions by ArganoCSS: Enterprise Performance Management

If you’re ready to gather real-time data and analyze predictive metrics for the success of your enterprise and its future, Oracle Enterprise Performance Management (EPM) Solutions can provide the technology you need. In today’s fast-moving, highly competitive business environment, your executive teams need an extremely reliable way to forecast, then monitor success, and Oracle EPM solutions provide you with capabilities needed for budgeting, planning, forecasting, modeling, reporting, analytics and more.

ArganoCSS has the proven experience needed to implement your Oracle EPM solution, and we recognize that no two enterprises are alike. This may especially be true when it comes to optimizing advanced forecasting processes and predictive analytics. That’s why we focus on your company’s specific needs and provide holistic solutions, and we excel in implementing systems that:

  • gather together disparate data
  • fill in the gaps
  • analyze inputs
  • provide predictive data to add visibility and value across your enterprise

We have extensive experience with Oracle’s EPM Cloud solution as well as its on-premise solution, Hyperion. Our highly experienced EPM consultants can help you determine the path that’s exactly right for you. To discuss customized solutions, please contact our EPM consultants online or call 1-800-814-7705. Let’s plan tomorrow together!