I have been long fascinated by the economics of cognitive work and I plan to explore this topic across a number of blog entries. I use the term cognitive work, not knowledge work, because knowledge work is a slippery slope. As soon as you know something, it stops being knowledge and soon becomes a fact. For example, when a person is scored for a credit card, that process of credit scoring used to be knowledge work, but because we have an algorithm for it, we don't think of it as being "knowledgeable" anymore.
For my purposes, I like to use the notion of cognition that people in psychology and artificial intelligence use. Cognition is a set of mental function, mental processes and states of intelligent entities which can be humans, human organizations and highly autonomous machines. (See the Wikipedia reference for cognition.) By this definition, the credit scoring process above would be cognitive work, both when the bank manager did it and when the credit scoring algorithm performed the task. This definition also allows for a cognitive framework which combines people and machines who may dynamically allocate decision rights.
I would also like to make two high level distinctions between Business to Consumer Businesses (B2C) and Business to Business Businesses (B2B). (It is possible for a firm to have B2C and B2B businesses within it. For example, The Travelers serves both individuals (B2C) and businesses (B2B)).
By and large, in B2C firms, the primary task is to automate and scale cognitive work. The core economics of a firm like GEICO are about codifying, automating and scaling decisions about customers, service, claims, payments and the like. When people talk about competing on analytics, this is usually what they are talking about. From one point of view, analytics are the "new science of winning". From another point of view, the data sets may be larger, and the sophistication of the models greater, and even the types of data entered into those models new and different (like telemetry data from cars, as one example of new data), but the notion of using data to routinize decisions was the core of Frederick Taylor's Scientific Management which was the backbone philosophy of the industrial revolution. The "new" part is that we are applying similar principles to cognitive work -- not just physical work.
The economics of these businesses usually involve a tremendous amount of automation and codification. Huge databases, call centers, web support and other tools of cognitive automation are central. Their marginal economics of serving a new customer are extremely low, and they are often driven by large volumes of transactions.
But when we turn the the issue of B2B work, we see an entirely different challenge. The core of B2B work is about coordination, collaboration and creative problem solving -- not codification and scaling. When an investment bank floats a new equity issue, they need to coordinate a tremendous number of participants inside and outside the bank, and the work cannot be sequentially broken down -- because it is a collaborative, evolving coordinating process. This is the core process in B2B work because one of the central distinctions of industrial marketing is that you have a complex buying group within the buying firm being serviced by a complex supplying group within the providing organization. Therefore, coordination and collaboration is baked into the very mode of value creation within the business.
In this realm of the B2B businesses, the core challenge for executives seeking to create a more productive organization is to create a shared information environment which can act as a platform for that innovation. For example, in one of our clients, they have created a single, integrated securities database which enables all parts of the bank -- from trading to treasury, to view all the positions of the bank. This enables them to do sophisticated modeling of risk, new product development, and effective proprietary trading. This common information space is a central part of what makes them more productive than many of their competitors.
In addition, B2B businesses need a clear architecture of participation. In this investment bank I mention above, the rules are very clear as to who has the authority to take which actions within the shared information space. It is essential that there is control over who can alter this common asset because people across the firm are counting on its accuracy. Put another way, in a B2C business if someone messes up a customer record, it will probably only effect that one customer. In the B2B firm, things are much more interdependent, and a mistake in the information space may propagate to other clients, and services rapidly. In the investment bank above, an incorrect record of a client's position can have implications not just for that client, but also for the hedging strategy, and funding needs at the very least.
The economics of B2B businesses are such that they spend a lot of money in communication and coordination infrastructure. They create cross-divisional and cross functional infrastructures that help the organization go to market in a coordinated manner. Often the tools they create custom tools to support their efforts because the value or coordination to them is so high that the nuances of their work must be reflected in their shared information environment. Their marginal cost to serve customers is usually relatively high, but their prices can be high and margins can often be greater than in the B2C firms.
In short, cognitive work and the related economics differs greatly in B2C firms and B2B firms. If executives try to apply the logic from one type to the needs of another, then there will be lots of work, and not much value. However, those firms that do get it right, create superior economics for their firms.
In future posts, I'll examine more examples, and some early economics given these ideas.
- Will Knowledge Work Ever Escape the Grip of Frederick Taylor? (A post I did last year.)
- Cognitive Reapportionment: A paper I did with Benn Konsynski on the implications of the ability to dynamically allocate decision rights between people and machines.