“Markets don’t fail. They always allocate goods and services perfectly,” a correspondent explained in response to last week’s column about market failure and how its sources might apply to IT shops that act as independent businesses selling to “internal customers.”

My correspondent’s position was, however, unassailable: Markets do always allocate goods and services perfectly, so long as your definition of “perfect” is “whatever allocations the marketplace delivers.”

With non-circular definitions, though, market failures are very real. Take my favorite example, covered in this space almost a decade ago:

The Dollar Auction

I auction off a dollar. There’s just one change to the usual high-bidder wins auction rules: The runner-up has to pay me his last bid too. So if the winning bid is a nickel and the runner up stopped at 4 cents, the winner would net 95 cents, I’d get 5+4=9 cents, and the runner up would be out 4 cents.

Except the runner up wouldn’t stop. He’d certainly bid 6 cents instead of losing the auction. And so on, and so on, until the high bid is a buck and the next highest is 99 cents. The second highest bidder now has to either bid $1.01 for my dollar — losing a penny on the deal — or stop, losing 99 cents. As losing a penny is better than losing 99 cents, there’s no end to the escalation.

Which, as I wrote in 2007, looks a lot like a politically high-stakes project that’s going off the rails. The sponsor can either throw good money after bad or cut her losses. But as each click of the throw-more-money-at-it ratchet looks to have a better ROI than losing everything invested thus far, and also avoids the political embarrassment of backing a loser, the train wreck continues into the indefinite future.

Customer Incongruence

The term “customer” involves three very different roles: Decision-maker (true customer), consumer, and wallet. When the same person fills all three roles, call it customer congruence, and market forces do what they’re supposed to do.

Customer incongruence (my term) happens when different people occupy the different customer roles, as when the family goes to McDonald’s for a Happy Meal. Be honest. It’s the kids who made the buying decision. Consumers? That’s the whole family. Mom or Dad are merely the wallet.

Healthcare provides another example of customer incongruence. The patient is the consumer, the insurance company is the wallet once co-pays and deductibles have been left behind. But who makes the buying decisions for your health care? For most of us it’s our doctor, who tells us what drugs to take and what surgeries to undergo.

That’s right: The seller of healthcare services holds the most important customer role: decision-maker. Which often leads to such inconveniences as buying a very expensive pharmaceutical solution when a relatively inexpensive alternative would be just as efficacious.

Which is not to suggest we should all prescribe our own treatments. Even with WebMD, few of us know enough. So while neither physicians nor IBM’s Watson are perfect diagnosticians, the alternative — self-diagnosing and self-prescribing patients — would result in a lot of unnecessarily dead people.

It’s a customer incongruity, which explains, at least in part, why the U.S. healthcare system is such a mess.

But IT organizations that act as a sellers to internal customers create customer incongruities that are just as challenging. The parallel with healthcare providers is, I hope, clear: Business executives and managers know where it hurts, but selecting or building IT solutions is complex enough to require professionals who know the field. They have the needed expertise.

As a result, to a very real extent, the IT organization acts as both seller and buyer of the company’s portfolio of information technologies.

The [partial] solution involves both formal governance and informal relationship management: Governance in the form of an IT Steering Committee or the equivalent — business managers who acquire enough expertise to oversee decisions about the company’s IT investments; relationship management in the form of the same trust-building you engage in with your doctor (and vice versa), and for the same reasons.

The better, admittedly partial solution is to not consider the rest of the business to be IT’s customer in the first place. It doesn’t do much for dollar-auction situations — what’s needed there is an executive culture that makes risk-taking safer by accepting that risk means some efforts must fail to pan out.

But customer incongruities go away when IT has no customers — when IT and everyone in the business collaborate to figure out and implement desired business changes.

I don’t know if it’s good economic theory. My experience, though, tells me it works quite well.

A couple of decades ago, when I joined Perot Systems, I greatly admired the weekly employee newsletter.

It was entirely prosaic — a text-only email layout-and-design tragedy where bold-face and italicized letters were the boundaries of formatting sophistication.

It was concise and readable, told employees the essentials of what was going on, and, every edition included a story. Not a tale of Ross and how brilliant and fabulous he was, but of a team that had done something that exemplified the company’s aspirations, goals, and values.

Which was far more effective in making sure all employees understood what constituted “how we do things around here” than any mission statement posters, values cards, or other empty gestures.

Perot Systems wasn’t a cognitive enterprise, but its employee newsletter was a step on the right path to making sure whoever made decisions on behalf of the company made decisions Ross Senior and his inner circle would agree with.

It’s no small challenge, and the bigger the company, the harder it is to push and prod the organization so it acts like an organism — a single entity with a single purpose. Large enterprises tend to be more like ecosystems than organisms. Why? Like ecosystems, the component parts of an organization are diverse, self-interested individual organisms — those pesky human beings you’ve probably had to work with once or twice in the course of your career.

Just in case you still aren’t convinced: Your brain, stomach, kidneys, and spleen all have different functions, but they all have the same goal — your survival and success.

Your company’s supply chain, IT, accounting, and manufacturing departments also have different functions, but there’s no reason to assume their managers and staff even care about the survival and success of anything beyond their little silo, let alone agree in any way, shape or form on how to achieve the survival and success of the enterprise as a whole.

One place to start is the golden rule of design: Form follows function, which is to say, understand the problem you’re trying to solve before you start designing solutions.

With a cognitive enterprise, one of the problems you’re trying to solve is how to give customers the impression they’re interacting with the equivalent of a person that acts with intention, not the complex, hard-to-navigate bureaucracy that’s the underlying reality.

There’s no single magic bullet for this. Creating a cognitive enterprise is a tough, tough challenge, as Scott and I discovered while writing the book. One starting point among many: Design the default sequences through which you expect typical customers to pass, and the mechanisms for exiting the default sequences when they don’t fit the situation.

Doing so enumerates what the customer touchpoints are. The next step is deciding what the customer experience should be within each one, for each channel through which customers can interact.

There are, as you might imagine, quite a few different variables to take into account. For example: Should you maximize the number of required touchpoints so as to create a soft, we-care-about-you impression, or should you minimize them so you don’t waste your customers’ time?

The answer, and you knew this was coming: It depends.

For example: If you’re an Emerald Club member and rent a car from National you can walk right past everyone, grab a car, and drive out. National caters to customers who appreciate convenience.

Enterprise, on the other hand, which is part of the same car-rental conglomerate, takes the opposite approach: Someone accompanies you to your car, walking you through every step of the rental process up to and including looking for dings and dents.

Enterprise figures its customers want the personal touch.

Which company is right? They both are. Different kinds of customer have different preferences. What the two companies have in common: Both started with what they wanted their customers to experience, then designed their processes and systems … and educated their employees … to provide that experience.

What else? Three rules:

  • If something interrupts the flow, what changes is the touchpoint sequence, not the touchpoints themselves.
  • Touchpoints are functionally identical, regardless of the channel. What customers do doesn’t depend on whether they’re using the phone, the web, or a mobile app.
  • Touchpoints might initiate or rely on back-office processes, but they do everything possible to hide those processes so customers don’t know anything about them.

Once you escape the one-dimensional mindset that everything is about cutting costs, creating the appearance of cognition really isn’t all that complicated.

In principle, that is. Making all this work in practice is as difficult as business gets.