Technology and the Right Decision Rights Decisions

My current print version of Information Week has an interesting cover story about “demand-signal management,” a highfalutin term for getting data about what’s selling, then using it to make decisions about manufacturing, replenishment, etc. The data come from the points of sale —  all the cash registers in all the all the stores in all the towns in all the world where products of interest are being sold —  and are voluminous. Think how many separate SKUs a supermarket or department store sells in a day, then think how many locations a large supermarket or department store chain has. And think how many days there are in a planning horizon. Finally, think of how many different supermarket or department store chains a large supplier deals with.

Old-school forecasting methods ignored this flood of data and based their projections on order and shipment data (rather than sales data). Demand-signal management (DSM) holds out the promise of improving results — better matching supply with demand in a fast-changing world — by incorporating sales information (which is, let’s face it, the most accurate signal about what’s selling) into forecasting algorithms and processes.

I started reading the article because of my geek interests in IT and supply chain management, but stayed interested because of a much broader interest (also geeky) in technology’s  influence on the design of organizations.

One of the most fundamental organizational design issues is the allocation of decision rights — in other words, answering the question “Who should get to make which decisions here?”  Some organizations are highly centralized; most important decisions are made at the top, and everyone else just executes on them. Many franchise restaurants work this way — headquarters decides on menu items and prices; the layouts, equipment, and technology for each location; the locations themselves; and most other important aspects of the business. Local managers and staff have relatively little freedom by design.

Other organizations are heavily decentralized, giving great autonomy to people far from headquarters. My brother used to work for Population Services International (PSI), a nonprofit devoted to improving public health. He spent several years in charge of their programs in Madagascar, and he had almost total freedom to decide what those programs should be. He was constrained only by his imagination, initiative, and funding; the PSI base in Washington didn’t want to dictate terms or tie his hands.

PSI took this approach because they believed that the most important knowledge about how to improve health in a given country could only be obtained by immersion in that country. As I wrote a while back, knowledge can be divided into two types: general knowledge (GK), which is easy to describe digitally and zip around the world, and specific knowledge (SK), which is hard to get out of people’s heads or codify at all. For any particular decision, the relevant knowledge for making it correctly is the sum of the general and specific knowledge about the situation.

PSI believes that specific knowledge overwhelms general knowledge in its sector: public health in the developing world. Starbucks believes, in contrast, that in its industry the general knowledge it possesses about how to locate a shop and make lattes dominates the specific knowledge of local managers and employees.

One of the basic trends I’ve been discussing with my MBA class this semester is IT’s power to convert specific knowledge into general knowledge. Companies today gather huge amounts of data over time and analyze it. This work converts qualitative intuition and experience (SK) into quantitative results and conclusions (GK). In addition, Enterprise 2.0 tools when they’re working well do a great job of harnessing and displaying collective intelligence. This is a crystal-clear transfer of specific knowledge into general knowledge.

So as I was reading along about demand-signal management I thought I knew where the IWeek article was heading. It would wind up by talking about how smart companies were centralizing their forecasting and planning activities to take advantage of the mass of point of sale data, and also centralizing subsequent decisions about replenishment, pricing, promotions, shipments, etc. I thought an article about a significant increase in general knowledge for retailers and their suppliers would also be an article about how these businesses were becoming more centralized.

Nope. The companies profiled, including Coty Fragrance, Best Buy, Goodyear, and Food Lion appeared to have a strong preference for having decision rights low in their organizations in the wake of DSM.  According to the article, for example, at Goodyear:

“More than 100 channel, brand, and category managers use the data for forecasting and planning, and sales reps access basic reports…”

At first I thought these companies were missing something. Then I realized I was. In fact, I was forgetting an excellent discussion we’d had in class about whether salespeople were surrendering a huge amount if they entered details about customer contacts into their company’s CRM system. This is clearly a conversion of SK to GK. Before CRM, only the salespeople themselves knew where exactly where they went, who they talked to, and what the substance and results of the conversation were. If they enter contact data into the system their bosses, their colleagues, and their replacements all have access to this information.

Some students in class argued that whatever the benefits of CRM might be to salespeople, they couldn’t outweigh the fact that the system turned their SK into GK. Others, though, maintained that even though this SK –> GK conversion did take place, it wasn’t that big a deal for two reasons. First, CRM didn’t and couldn’t come close to capturing all a salesperson’s SK, or even the most important bits of it. These people amass knowledge about customers and their companies —  who’s amenable to a hard sell and who’s not, who seems to have stopped paying their bills, who’s about to jump ship to a competitor, what kind of body language works for each person, etc. —  that’s hard to capture in a database, and often even hard to articulate at all.

Second, knowing things is only part of the job for most knowledge workers. Another big part is communicating effectively with other people, and communication is a subtle art. I could study the transcripts of a master salesperon’s interactions with customers to the point that I’d know what words to say, but I assure you that I’d still be a horrible salesperson. In their excellent book The New Division of Labor Frank Levy and Richard J. Murnane describe how ‘complex communication’ is one of the things that people just do better than computers, and so one of the things that will prevent machines from taking over all the jobs any time soon.

As it turns out, effective supply chain management involves a great deal of both specific knowledge and complex communication, even after DSM is in place. As the IWeek article states,

“Demand-signal analysis may provide earlier and more granular insights, but the problems manufacturers and retailers are trying to solve have been around for a long time. It’s essential to have people steeped in industry knowledge leading this effort.”

The companies described in the article are thus making a very smart choice by being decentralized in this regard, even after snazzy new technology becomes available. Examples like this one educate me, and make me more cautious about assuming any broad trend toward either centralization or decentralization as we continue to inject more and more technology into every industry in the economy.

I’m confident that because of IT we’re in the middle of a period of broad and deep exploration about the ‘best’ design for a given organization, and a time of increased experimentation with allocating decision rights. And I’m not sure about where we’ll end up.

Are you?  Do you think technology is generally leading us to either greater centralization or decentralization of organizations?  Leave a comment, please, and tell us what you think and why you think so.