Research

Leave your comfort zone, get more from EPM

Companies have invested heavily in information technology known as Enterprise Performance Management (EPM) systems, which generate everything from basic budget and financial statements to complex forecasts of how to best meet consumer demand. But too few firms are using the systems to their full potential, according to Professor Robert St. Louis and doctoral student Jeremy Glassman.

2013-05-31

Companies have invested heavily in information technology known as Enterprise Performance Management (EPM) systems, which generate everything from basic budget and financial statements to complex forecasts of how to best meet consumer demand.

But too few firms are using the systems to their full potential, according to Information Systems Professor Robert St. Louis and doctoral student Jeremy Glassman. Their recent research offers clues as to why: It's not the technology of EPMs, but good old human nature. Business leaders are less than comfortable with non-financial data and with the forecasts based on such data. But ultimately, the researchers say, businesses must take risks today in order to realize EPM's benefits in the future.

Humans tend to get uncomfortable when we acquire sets of information that are highly inconsistent with each other -- an experience known as high cognitive dissonance. To make the information sets more consistent and make ourselves more comfortable, we tend to discount or even ignore the less credible information. Many companies and business leaders regard financial data as more credible than non-financial data, so putting less value on the non-financial data restores their comfort levels to familiar territory.

In a rapidly changing world, though, comfort won't do.

Strategic advantage ahead

"There's a transition, I think, that companies need to make," said St. Louis. "They're used to using information systems to generate reports and to generate financial statements, and those are all backward-facing and historical. ... If people are really going to benefit from information systems, they have to start being forward-looking instead of backward-looking."

Glassman, a Ph.D. student in information systems, said the issue of under-utilized EPM systems arises when companies first implement their systems. "They know there are questions to be asked, and they know they need a system like this to ask those questions, but they don't know what the questions are," he said.

Profitability modeling and optimization are the two most useful forward-looking components of EPM systems, St. Louis said, and those are the components that offer businesses the best shot at gaining strategic advantages.

But managers, doubting the credibility of non-financial sources and forecasts, tend to trust their instincts more than those helpful components of their EPMs. Specifically, St. Louis said, there are two big barriers to fully using EPMs:

• From the mountains of data available, managers must decide what is relevant to a decision and what is irrelevant, then build a model to give them control of results. Still, they lack confidence that the model really will work.
• Even if managers develop a model that seems to work, they lack confidence that they are using it correctly.

Models, however, are key to a company’s future success. "They have to go from this backward-looking report-generation mentality to this forward-looking predictive mentality, and that's just impossible to do without using models," St. Louis said.

The forecast for fruit

For Sun West Fruit Co., a California company that grows, packs and ships fruit, learning to deal with non-financial data was a necessity. Glassman had worked as a consultant to the company and observed successful use of EPM first-hand.

The company uses financial data from its sales and distribution systems to measure changing demand for different varieties of fruit. It then consolidates its financial metrics with non-financial ones, including weather forecasts, economic projections and environmental variables, to forecast changes in the supply of and demand for its products.

Those forecasts don't come easily. Sun West has to estimate demand for its product mix of multiple varieties of oranges, peaches, plums and nectarines two to five years into the future, because it must decide what trees to plant and how many to plant to meet demand, then wait four or more years for the trees to mature. It also has to plan for the type of packing equipment and amount of storage space it will need to meet the demand.

"In their industry, what is ripe is constantly changing," Glassman said. "Their customers just want a peach, plum, nectarine or orange, and Sun West has to figure out how to match that demand with one of 150 varieties based on a certain level of ripeness, certain qualities and a set price point."

The company also uses profitability modeling and optimization to calculate the cost of sorting fruit by qualities such as ripeness, appearance or sugar content; then to calculate the prices it can charge for such qualities; and to devise a product mix that optimizes profitability.

Sun West also uses the strategic management component of its EPM to help field crews and sales people communicate which products are ready to harvest and the quantities needed at specific times to meet orders. The system also communicates the schedule to the processing plant, so it can have staff and machinery ready.

Being part of the agriculture industry, the company also has to pay attention to long-term weather forecasts and averages, which affect the trees' yield. "They had to get more involved in the planning and forecasting, and also out of necessity, they had to get more involved in the profitability modeling and the optimization," St. Louis said. "They really couldn't be effective without doing this."

Asking the right questions

The company started its forecasting process about 10 years ago using Excel spreadsheets, Glassman said, then automated the process with an EPM system. Those years of experience with its models gave Sun West insights that other EPM users have yet to uncover about their businesses. "To ask the right question of an EPM, you first have to be able to infer what you think the right answer is or at least what the question could be," he said.

Once Sun West managers started using such data, it helped build their confidence that their model worked and that they were using it correctly, St. Louis said.

Other industries might not worry about long-range weather or crop yields, but they almost always have to factor some sort of non-financial data into their plans. They might look at trends in consumer tastes or predictions of what consumers will value, St. Louis said. For example, most companies will have to consider changes in technology, such as the ability to take pictures of checks and deposit them in a bank account, or how quickly consumers will adopt new technology, such as the massive switch to digital photography from traditional film cameras.

When considering non-financial data, managers also have to think about managing risk. "If you follow your metrics, sometimes you're going to forecast wrong, but if you see you're not meeting your goals, you have to figure out why and make adjustments quickly," St. Louis said. EPMs give companies that flexibility, but historical reports cannot, he said.

The companies that will do well in the future are the ones that will spot trends early, making it even more important to fully use their EPM systems, he said.

Ignoring an EPM's models is much easier than putting in the hard work of finding accurate data and using the models correctly, he acknowledges. To get companies on board, proponents of forecasting and optimization will have to make future benefits more real, or salient. The two researchers suggest doing this through cognitive feed-forward -- well-defined forecasting procedures -- and cognitive feedback -- comparisons of actual and forecasted demand. Sun West, they said, believes the combination of both strategies accounts for managers' acceptance of its models.

To illustrate a way of making future benefits salient, St. Louis cites a model used in a study of people trying to lose weight. The study subjects were told that if they ate a certain amount of calories and exercised a certain amount, they would lose so many pounds in the first month, so many pounds the second month, and so on. People who followed the model saw the forecasts come true quickly, and that made the future benefits of the prescribed diet and exercise appear more real to them.

With companies, it is not as easy to develop a model and prove within a month that it works, St. Louis said. But if companies get used to using their models and tracking the results, they will gain confidence that it works.

Uncertainty about the credibility of the source of data going into a model is another barrier to fuller use of EPMs. People have to have confidence in both the accuracy of the data and their ability to forecast accurately with it, St. Louis said.

Sun West looks daily at its forecasts of demand and sales. The forecasts are important to its profitability, because it is more expensive to pick, pack and ship fruit that will go unsold than it is to leave the fruit on the trees. The demand data tells the company how much fruit to pick, pack and ship. It now has a track record for its forecasts, giving it confidence in the models. "It's really one of those cases where success breeds success," St. Louis said. "They saw that using these models in trying to forecast consumer demand worked a lot better than just trying to use their intuition of what the demand ought to be."

Lessons learned

There are two lessons for those who want to put St. Louis and Glassman's research into practice and get the most benefit from EPMs. First, be aware of your company's own cognitive dissonance, or comfort levels with its financial and non-financial data. Second, look for ways to increase the credibility of the non-financial data and of models driven by non-financial data.

St.Louis acknowledges that getting companies to use and trust their models is a chicken-and-egg problem. If they never take the steps to use it, they'll never experience the results. And even if they do try modeling, it might not forecast accurately the first time, prompting a human tendency to scrap the model. Instead, managers should adjust, refine and build a better model.

Glassman said the researchers recommend that companies understand where their data, both financial and non-financial, comes from, which will help make their decisions and analyses more valuable.

St. Louis is heartened that more companies are starting to use the forecast and optimization components of EPMs. And there are always people who like to try new things, and others who get on board when they see the new process work.

"We recommend they really start trying to fully utilize both the planning and forecasting components of the models," St. Louis said. "We know there is going to be tremendous change coming in the next few years, so that makes this ability to plan, forecast and optimize be much more important than it has ever been. Yes, there is some risk associated, but there's also a greater payoff."

Information Economics and Strategy IT-Enabled Services and Process Innovations Managing IT