Don’t Listen to Jack Welch (Only His Best Part)

Jack Welch advises executives to leave their rooms and find out more about organizations they manage.

I’m afraid, this is what executives will do. Why? Welch makes two points. First, he shows the problem, which is real for sure. Knowing your organization is important. Second, he suggests to solve it by visiting “stores, trading floors, regional offices, factories.” It’s also a good point, but not the best solution.

By taking Welch’s advice literally, executives will find no more than a mess of emotions, stories, suggestions, and demands. It’s like reading a morning newspaper: you really need a lot of prejudices to make sense out of this flow of information, when this flow doesn’t have any sense. It’s best at confirming existing prejudices. If you really want to know something about the world, you should do a comprehensive study on topic.

How does it look in management? If you want knowledge, organize it. Build an IT system that let your people talk freely (even if anonymously), send requests to supervisors, get feedbacks, and discuss ideas in a single place. Not face-to-face meetings of the king and His Majesty’s subjects (it always looks this way). It must be a distant platform. A person must know it’s for real and feel no pressure.

Computers are stupid but extraordinarily good at handling whatever comes out of this. Can a human delegate 8.5 million problems to 3.7 million solvers in milliseconds? StackOverflow does this routinely and arguably saved more working hours than YouTube wasted. It’s a matter of minutes to find popular problems, topics, and experts. It’s easy to find where your help is needed. This system shows what matters.

You can spend time traveling around “stores, trading floors, regional offices, factories” to declare, like Jonny Cash, “I’ve been everywhere.” Or you can systematically improve the system that delivers real information from real people right to your armchair. An IT system is better at everything that travels can do: moods, relevant problems, upcoming disasters, and best ideas. Exciting travels, as Welch noted, show that you’re not alone. But they are not for decision making.

Computer-driven operations at Amazon and Walmart have beaten flesh-and-blood shops around the corner. These systems know what customers want, unlike shopkeepers who talk to their customers for hours each day. There must be some sense of modesty regarding own abilities to admit this, but it would be one level up in business management. The creators of Amazon and Walmart could improve because they recognized their limitations and let machines do their work.

This transformation is slow in management because of the email reputation IT systems have. They’re something delivering tons of letters you have no time to read. It’s a failure of design. Emails came from the 70s and haven’t changed since then. ERM and other “management” systems often copy emails in asking too much irrelevant information. They lack human input and the sense of importance. But that’s how public web services looked in the 1990s. Since then they’ve changed tremendously; and so will B2B systems. Don’t miss this moment traveling.

Talent Shortage or Management Problems?

Russian soldiers cleaning a backyard. Talent shortage or management problems? Source

A famous quote attributed to Mark Twain advises, “Buy land, they’re not making it anymore.” That’s how executives think about employees.

Talent shortage tops other strategic issues in executive surveys. When a TV remote starts looking like an airplane cockpit, employees need skills to do their jobs. Companies go to the job market after these employees and find the shortage instead. It’s easy to find this shortage. Just offer a salary low enough.

Another shortage is the real one, when companies pay high wages to folks with skills. Who has this shortage number two? The US does:


Since the 80s, college grads got paid more for their skills, compared to those without degrees. The recent decade looks kind of flat, and filling positions should be nothing new to managers. (And, no, the college premium isn’t about entry positions alone. It also indicates what’s happening with top job openings.)

Other countries, though, should experience problems with finding the shortage, not with finding employees:


Premia are flat everywhere, but in the UK and Germany, for 15 years. Latin America:


Brazil does have a skill deficit and high returns to schooling, which mean the economy really needs more educated workers and firms are ready to pay for them. The rest of the countries have other problems to solve before setting to with skills.

The talent shortage seems serious when productivity differences within countries are high. A firm knows that competitors have skilled workers and succeed, while its own resources and performance are depressing. Since all other important factors, like management, are unobserved, it’s tempting to conclude that best workers create great performance. Partly they do, but this thinking solves nothing. A poorly performing company can overpay for talents, but under weak management, increasing costs offset the small gain in labor force quality.

Money and people are in deficit because of the steam engine thinking: the more coal you throw into a furnace, the faster you go. Though that’s technically true, replacing the engine is a better idea.

The number of firms and magic numbers

In business literature you can find claims about magic numbers in management. The number of ten is said to be where a firm starts, while anything below that is a team and needs a different approach. One hundred employees are the milestone where the CEO no longer knows her employees personally and corporate politics emerges. And so on.

Management depends on the size, of course. But the data points at no specific critical numbers. The histogram of firms by the number of employees:

(hat tip to Jeffrey Groen at the BLS for the data) 

And the log scale:

If ten or any other number had any significance, we’d see abnormal behavior around it. Management experiences transitional difficulties and some firms disappear. It would look like discontinuities or multiple peaks, as with the middle income trap. But the distribution does not support this claim. Perhaps, specific numbers matters for certain subpopulations of firms, for instance, the IT industry. In that case, it’d be interesting to look at the data. But until that it would be overall great if authors were more responsible and back their statements with minimal evidences.