In a recent post to the Big Fat Finance Blog, BPM guru Gary Cokins argued that a tectonic shift is under way in accounting as companies transition from a decision-making approach based on descriptive, historical data to a forward-looking, participatory style of management. Cokins, product marketing manager with business analytics and BPM firm SAS and author of several books on performance management and cost accounting, talked with BPM about the hurdles companies face and how performance management software can help.

BPM: What’s wrong with management accounting as practiced today?
Cokins: The broader problem involves why financial accounting dominates managerial accounting to the degree that organizations can’t give sufficient attention to improving the shortcomings of their managerial accounting system.

When you step back a bit and look at what’s important, financial accounting simply deals with valuation — for example, what would an organization be worth if you were to sell it? It deals with compliance and regulatory reporting for the external investment community and governments. But managerial accounting is about creating value — its information contributes to management decisions internal to an organization that financial accounting ultimately deals with afterwards. Which type of accounting is more important?

I realize it’s important to the capital markets and the financial investment community to have confidence in financial reporting. But I think it’s even more important for managers and employee teams to have reasonably accurate and relevant managerial accounting information to make better decisions with.

Managerial accounting is one the important components of the various methodologies of the business performance management (BPM) framework. Its information is what can give an organization a superior and sustainable advantage to satisfy the needs of its investors and employees.

BPM: How can BPM systems add value to cost data?
Cokins: I’d reverse the order and suggest that cost data adds value to enterprise performance management systems. Part of the problem is that there’s confusion in the marketplace about the term “enterprise performance management.” EPM is often perceived as just a bunch of measurement dashboards for feedback and better financial reporting. But it’s much, much more.

The good news is, performance management is not a new methodology that everyone now has to learn — rather, it tightly integrates business improvement and analytic methodologies that executives and employee teams are already familiar with. Think of performance management as an umbrella concept. It integrates operational and financial information into a single decision-support and planning framework. Ideally it embeds each methodology with analytics of all flavors, and especially predictive analytics.

The methodologies within the performance management framework include strategy mapping, balanced scorecards, costing (including activity-based cost management), budgeting, forecasting, and resource capacity requirements planning. These methodologies fuel other core solutions such as customer relationship management (CRM), supply chain management (SCM), risk management, and human capital management systems, as well as lean management and Six Sigma initiatives. It’s quite a stew, but they all blend together.

Some components of EPM provide useful information without requiring cost data. An example would be how marketing and sales departments analyze which types of customers are attractive to retain, to grow, to win back, or to acquire. However, their interest is mainly to increase sales and market share. By adding cost data, you can determine how profitable a customer is today, and you can also look forward by calculating customer lifetime value (CLV), which assists in optimizing how much to spend on each customer or their micro-segment. It makes the goal to increase profitable sales — smart growth.

BPM: Given the sheer volume of information that companies generate, why is it such a challenge for them to use it proactively?
Cokins: Your observation is correct. Organizations are drowning in data but starving for information. Data and information are not the same thing. BPM and BI systems convert raw transactional data into information to be used for better decision-making.

An organization may be able to perform data mining, to drill down with query and reporting tools, but that’s not enough. Yes, it’s good to look in the rearview mirror and analyze historical information for patterns and insights. But decisions are always made in and for the future. One must look through the windshield. Most organizations are shifting from a command-and-control style of management — looking at historical reports and reacting — to a forward-looking, anticipatory style of management where they’re looking at the future coming at them and proactively making adjustments or resolving small problems before they become big ones.

The challenge organizations have in looking forward are insufficient skills and lack of capabilities with predictive analytics. The technical problems with their treasure troves of raw data are solvable with data management and integration tools like extract-transform-and-load (ETL) tools. Where organizations fall short is in achieving competencies with estimating and forecasting techniques, such as statistical and regression analysis. Success with this requires executive sponsorship to create a culture for metrics and quantitative analysis. Decision-makers can no longer rely on gut feel and intuition.

A recent survey by the consulting firm Accenture reported that most companies are far from where they want to be when it comes to implementing analytics; they’re still relying on gut feeling rather than hard data. They’re committed to improving, though. More than two-thirds of respondents said their organization's senior management is totally or highly committed to analytics and fact-based decision-making. My takeaway is that the challenge to apply predictive information will be confronted as it becomes a priority for executives.

BPM: How is the shift to predictive cost planning impacting the BPM market?
Cokins: It’s affecting two groups: one, companies whose traditional planning and budgeting methods are falling short; and two, the BPM software vendors who provide the technologies to enable predictive analytics.

Let’s start with the users, typically at first the CFO function that creates the planning information for the line managers. Increased volatility is now the new normal, and that includes consumer preferences, exchange rates, and commodity prices, to name just a few. Trends can develop quickly such as the emergence of competitors in developing economies like Brazil, and the rise of Web-based social communications. Unanticipated shocks can come from occurrences like the Asian tsunami, H1N1 flu, or the current global credit crisis. This means that traditional practices like detailed annual budgets and five-year plans can quickly become obsolete.

The shift is toward more agility, speed, frequency, and visibility in reporting of all information, not just managerial accounting information. More importantly, the reporting must more quickly help users to gain insights and inferences. Rather than the accounting department annually producing detailed line-item expense budgets, the shift is toward rolling financial forecasts with less detail and more summarized information as the planning horizon stretches into the future. The sources that are needed to calculate the level of resource capacity and spending come from a variety of other forecasts, such as demographics and sales plans. Predictive analytics are needed for each source, and they are all eventually aggregated in the financial projections.

Risk management can’t be left out of the equation. The potential magnitude of an unplanned and undesirable event and its likelihood of occurrence dictate how much risk mitigation and insurance-like projects are needed. These too are increasingly included in financial projections.

Now let’s turn to the BPM software vendors. They try to stay ahead of the curve. However, futurists like Thornton May believe that management is facing an inflection point where the application of analytics is becoming mission-critical as a basis for a competitive edge. Just look at the business model for Netflix. It’s based on an algorithm that gives customers suggestions about their next rental choices based on their rental patterns and similar movies.

So the impact on BPM software vendors is to push them to provide robust and scalable computing power with analytics of all flavors, such as segmentation or correlation analysis, embedded within all of the BPM methodologies, including balanced scorecard measurements and, obviously, financial projections. The software for the large ERP vendors was designed for transactional processes, not near-real-time analytics. There will be a significant re-tooling involved for them to support analytics.

BPM: Don’t companies already have a prospective view of costs in terms of what expenses are fixed, semi-fixed, and variable?
Cokins: The key point is that both financial and managerial reporting evolved primarily for historical reporting. That’s why they’re called “accounting.” I morbidly refer to them as the “profit and cost autopsy.” The revenues and expenses already occurred, so they are precisely known. The question is, which products, service lines, channels, and customers consumed the time period’s expenses? The principles of activity-based costing greatly improve the accuracy of the results, and as a bonus ABC provides visibility as to what drivers caused the results. But for the prospective view, the economics of resource capacity behavior comes into play.

To better understand the prospective view, one should consider the types of decisions that are made with predictive costs. They’re not restricted to budgets and rolling financial forecasts. They can include capital investment justifications, make-versus-buy decisions, outsourcing decisions, and what-if analysis. For these types of questions, as the planning horizon extends into the future, resource capacity can be adjusted; you add some and remove some. And so the classification of a resource — a machine or a type of labor skill, let’s say — can only be tagged as fixed, semi-fixed, variable, or discretionary, relative to future demand based on the question being asked.

Most companies don’t have a system designed for this type of economic analysis. Their accounting system was designed for reporting, not analysis. This is why various departments, such as engineering, create their own system to accommodate modeling fixed, semi-fixed, and variable expense behavior. In some cases, this type of analysis is a one-time study with much effort to set it up. However, as I said earlier, volatility is now the new normal, and it makes sense to have a permanent and repeatable managerial accounting system for the entire enterprise. Most managerial accounting and budgeting systems will need to be upgraded to reflect widely varying demand behavior, not just for forecast customer orders, but arguably for any type of economic analysis.


BPM: There’s an abundance of terminology in this area, and it can get confusing. What’s the difference between data mining, predictive modeling, and predictive analytics?
Cokins: Data mining was the buzzword about 10 years ago, but the terms predictive analytics and predictive modeling have become more popular recently. These terms are related, but they don’t cover the same activity.


Data mining has been defined in a lot of ways, but essentially it’s a process for analyzing data that typically includes steps such as formulating the problem, accumulating data, and training and evaluating models.


Predictive modeling is an analytics technique that answers questions such as: Who’s likely to respond to a marketing campaign? How much do first-time purchasers usually spend? Which customers are likely to be a credit default?


Predictive analytics is an umbrella term that encompasses both data mining and predictive modeling, as well as other analytical techniques. It’s a subset of analytics, which more broadly includes other areas of statistics like experimental design, time series forecasting, operations research and text analytics. Predictive analytics can be defined as a collection of statistics and data mining techniques that analyze data to make predictions about future events.


And then there’s forecasting, which is similar to predictive modeling, but with a twist. Here’s a quick analogy: Forecasts tell you how many ice cream cones will be sold in July, so you can set expectations for planned costs, profits, supply chain impacts and so on. Predictive models tell you the characteristics of ideal ice cream customers, the flavors they will choose, and coupon offers that will entice them.


If your goal is to do a better job of buying raw materials for the ice cream and to have them at the factory at the right time, your company needs a forecasting solution. If the marketing department is trying to figure out how and where to market the ice cream, it needs predictive modeling.