Google's achilles heel is service delivered by people

They don't scale people. (Statue of Achilles dying from Wikipedia. The original is at the Achillieon near Corfu.)

In some ways this makes perfect sense, because people don't scale well. There is only so much technology you can place in front of an "operator standing by." Once her calls are queued and she's in front of a terminal you're at the mercy of the callers.

People get sick. They have lives. They miss shifts. But the biggest variables are on the other side of the line. Some people can't make themselves understood. Some won't shut up. Some don't have what they need in front of them before they place a call. Some people get mad. Some people lie.

So anything that requires the use of many ordinary people, anything that's people-intensive rather than compute-intensive, Google avoids.  There is no Moore's Law of learning, or of training.

I was reminded of this over the weekend, when a good friend relayed the story of a client whom he'd encouraged to use GMail, Google Docs, and other Google services.

The client's account was cancelled, supposedly for violating Google's terms of service, and he couldn't get through to a person to settle things. A waltz through the valley of Google voice mail ended up in virtual jail -- he was told they just don't offer live support.

The story has a happy ending. The account was unblocked as mysteriously as it had been blocked. But this is not the only place where Google fails due to its faith in machines.

Google Health, which I have covered since its launch, is another example. Microsoft is beating Google's butt in the health IT business, because Microsoft hires enough people to contact the customers, understand their needs, and adapt to their whims.

As a result Microsoft Healthvault is gaining serious traction as a Personal Health Record (PHR) system, while Google trails. The Google booth at HIMSS 2010 was nearly identical to what I had seen at HIMSS 2008. Microsoft, by contrast, is now a big vendor, a major name, alongside HP and IBM, or such specialty firms as McKesson and Cerner.

Google has two options here. It can hire people. It can use its chat and GTalk systems to connect with customers, maybe charging for calls that don't go over its network, and it can provide real service. Google has operations around the world, this is not a difficult thing for it to do.

Or it can accept the problem and avoid those lines of business where you need a lot of people in order to make customers happy. That means dumping things like Google Docs. That means dropping the very idea of customer service because it's just too hard.

The folks at the Googleplex have put this decision off too long. What they have now are an increasing number of unhappy customers, and an increasing number of failed initiatives, because there aren't enough Googlers with average incomes.

Or it can put its talent to work becoming a jobs machine, and accept the fact that people are messy.