After years spent developing reliable data sources and reporting tools, management teams are now confronted with as fundamental a problem: decisions made on the basis of global data, mostly accurate, yet often grossly misleading when it does not reflect the behavior of underlying data points.
Let’s take a few concrete examples: First, a distributor decides to run a promotion with a 20% discount on all available products without any exception. Once the sale is over, the chief accountant reports to the Sales Director: “You have sold the same merchandise quantity as last years’ sale, yet the total average price has increased and so has our total revenue.” The Sales Director and other managers in the company are dubitative. How could average price increase when the price of all product items has decreased by 20%? Fortunately, the chief accountant can explain: Customers took advantage of the sale in order to buy overall higher priced products. So the global average price increased, despite the fact that all individual product prices were lowered.
Another example: Unions complain that a salary freeze has been in effect over the past three years. However, top management insists in good faith that the average salary has increased! Who is right? Everybody, as some of the manufacturing is now subcontracted in Asia. As a result, there are less factory workers on payroll and the proportion of white collar personnel has logically increased. So the company has gone toward a proportion of higher salaries, and mathematically the average overall company salary has increased, despite the fact that individual salaries haven’t.
Here is a third example: a company selling to a large retail chain has segmented its stores by regions. The company’s new Sales VP suggests that management time should focus on main regions, and that two smaller regions that contribute seemingly little to overall company revenue, should be left aside. But some among the sales force don’t agree: They put forward that one of the smaller regions, having indeed only a few stores, has the highest per store revenue of the whole chain. Plus those stores are fast growing. So what we have here is a small region on a global level, yet the best performing stores on a local level.
Those kinds of diverging perspectives between global and local, strategy and operations, are more the rule than the exception, contrarily to what one might believe. And these are not subjective perceptions but hard data. From that standpoint, we have introduced a new kind of metrics in order to measure disparities between global and local viewpoints: The “Intra share” (or simply “Intra”) measures the part of the local that behaves like the global; opposite, the “Extra share” (or simply “Extra”) measures the part of the local that doesn’t behave like the global. Thus the Extra share is a metric that enlightens potential differences between global and local behaviors. In reference to the above examples, the Extra share is each time equal to a 100%: A global average price increasing while all individual prices are decreasing; a global average salary increasing and individual salaries that do not; smallest revenue on the regional level with highest per store revenue. Thus three times in a row, we see global entities that behave entirely differently than their micro components.
We just went through a few basic examples. Yet, whatever the situation of an enterprise may be, our experience shows that Extra share is generally high. Without getting here into any detailed survey, let’ say that the Extra share is usually comprised between 30% and 60% with frequent picks of 70% and more. That really tells how different behaviors between global and local can be. How then do top managers run their company with management reports that often yield a very distorted view of operational realities?
At the level of a management team, such issues remain mostly unaddressed except in particular problem cases. Top managers rarely have the time to consider more granular or detailed data. Fortunately, managers at large know their business and their data (as in the example above regarding the store chain). It is human intervention on all enterprise levels, that makes up for present limitations of reporting tools.
Some believe, however, that such limitations are immaterial, and that they can be easily overcome with a needs analysis, in order to determine the most salient and transparent metrics to be used. Thus in the above retail chain example, one could introduce “per store revenue”, a metric that would provide a local view of the revenue mix, not only a global one per regions. But one needs to know ahead of time that per store revenue is relevant, which would not have been the case if stores were all similar in terms of revenue. Let’s take yet a third example: total company margin has increased 5%, however 70% of revenue comes from products whose margin has declined! In order to account for that remarkable phenomenon, due to mix variations across the product line, one can introduce a metric tracking each product contribution to global margin. But there too, one needs to know ahead of time the nature of the problem in order to devise a relevant explanatory metric.
Neither industry knowledge, nor the best needs analysis can aspire to uncover all data discrepancies, especially when performance may vary unexpectedly and when complexity is an issue: what to expect from a large corporate entity or even from a medium size business with thousands of customers and products to manage? Or from fast moving markets with many competing products? What should one do when it gets difficult to dwell into detailed data without loosing perspective? As the aphorism goes, “the devil is in the details” and may invalidate the best strategy, thought out from a global perspective.
Moreover, appropriate metrics change over time, given moving company environments and strategy reassessments. The “agile” enterprise needs “agile” metrics. Extra share (see above) allows pinpointing immediately divergences between global and local viewpoints, between strategy and operations… and may also be a basis to introduce, if needed, new explanatory metrics.
A first stage in order to structure that approach, would be to introduce the Extra share metric in company reports, allowing to know where to drill-down into detailed data. If Extra share is small, underlying data is rather homogeneous and global is a good approximation of local. However, when Extra share is large, underlying data are heterogeneous, and global is probably a poor approximation of local. There it is key to dig into details. A second stage would be to use complexity exploration tools, enabling to better manage unavoidable divergences of views between organizational levels. One more great undertaking for the future of Business Intelligence, to which EFFIS is already contributing with its working methodologies and proprietary software.
No comments:
Post a Comment