“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.