Management
In 2016, I was a candidate to be dean of Stanford Business School. I’m the author of two business books, one of which was on the BusinessWeek bestseller list.
Here are a few thoughts on management and business …
Does management work?
It’s not an easy question. Certainly, some businesses are well managed. But the business press is full of fashionable gurus promoting the latest management fads. An excellent article shows that people quit jobs, not firms. The shift from top-down to flatter organizationsprobably hasn’t really done much to improve employee engagement or effectiveness. Self-management is an ongoing experiment. There are many management systems and anecdotes of success, yet those researchers who study failure find the same principles at work that are supposed to cause success. Some companies have found the secret recipe or developed systems that work, while others are still working on it and most others have not. If you think people work for money, watch Daniel Pink unpack the latest research:
Do boards of directors help a company? Probably not, though few researchers have studied this. CEOs choose board members who are aligned with their salary goals. Many boards are passive. They rarely prevent disaster, though some might. Studies point to long-term stock ownership rather than short-term compensation as part of the solution.
Is CEO pay justified? Here, the data is mixed. On one side, we have data showing there is no correlation between CEO pay and company performance.* A growing body of research also shows that pay for performance does not work, yet CEO salaries drift ever upward as a result of popular myths that these tall, mostly white male, lucky people are worthy leaders. On the other hand, one study showed that companies with well-compensated CEOs far outperform those with lower-paid CEOs and Tyler Cowen argues that CEOs capture less of the value they bring to corporations than other employees.* Correlation or causation? I guess it’s a little more complicated than I thought.
How much should we trust experts? Not much. In engineering, they mostly agree. But in complex adaptive systems, there are many experts with opposing views. David Freedman has shown that experts are overconfident and often wrong. The more “important” they are, the less we look at their actual track records. Double-blind taste tests show that wine experts cannot consistently identify wines they have previously identified many times. Economists are worse.
The Value of Strategy
How do they do strategy at Google? They don't. Because Google employees spend 20% of their time doing anything they want, they are always trying hundreds of experiments. Probably no company fails as much as Google does. Once a project has momentum, volunteers swarm toward it and focus on execution. Google's products always come out half-baked and can stay in that condition for years as they iterate and scale. Yet Google’s products continue to improve every single day. Google is a company for the 21st century, because they don’t try to predict the future; they just try things and see what works.
More and more companies are seeing the wisdom of replacing strategy with experiments. A good experiment is cheap to run and gives you a signal to follow (or dies, providing valuable information). If you can do ten experiments and get one strong signal, you may find it's worth the expense. Venture capitalist Mark Andreessen has gone as far as predicting the death of physical retail stores, because online retailers can do A/B testing so much more quickly and effectively. Experiments are often small and fast and you can get the results in a matter of days, not weeks.
Experiments should come from many different directions. You should expect to look at several of them and wonder whether they could possibly work at all. Recently, a web site called WeeSpring discovered that when they forced people to register before visiting the site, more people registered and visited than when they made it optional. This was a surprising result, one they could only learn by trying something that looked like a mistake. In general, rather than answering a question with a prediction, it's better to say "I don't know, how can we try it?"
Randomness can play an important role in experiments, especially when the cost of experiments is low and the payoff of a win is high. Bike manufacturer BMC employed a software program that literally generated thousands of random possible designs, tested them all, and showed the winners - automatically. They called this "accelerated evolution." It led to the development of new frame features they wouldn't have produced by asking the design team to draw up their best ideas. This kind of random trial-and-error process using software to test and promote successful candidates has been used in drug and compound discovery for decades now.
It's very likely that experiments can enhance, if not replace, almost all of your planning and strategy. It's easy to think of big strategies that have entered the market stillborn, from soft drinks to cars to advertisements to many exciting new consumer brands that just didn't take. Why were the companies taken by surprise? Because they ran focus groups that validated the executives' decisions, rather than trying it in the market and seeing what happened.
Cognitive Diversity
Kevin Dunbar’s research at the University of Toronto has shown that teams of people with different backgrounds and viewpoints solve problems faster and generate more ideas to choose from than teams of similar people. Yet teams are often assembled by the same person or process, building teams of people who are too similar.
At Google, for example, new potential employees are interviewed by a number of different people, even those they wouldn't be working with directly; then these interviewers tell another group (who haven't met the candidate) about their impressions and the second group makes the hiring decision. This avoids both groupthink and hiring people current employees like, which has no correlation with job success.
Philip Tetlock says a “hedgehog” is a person with a single big idea who drives his agenda forward relentlessly. This describes many managers, executives, and pundits. A “fox” is someone who has many small ideas and changes her viewpoint as she gets new information. Tetlock’s work on prediction has shown that foxes consistently outperform hedgehogs in predicting future events. Since key executives tend to be hedgehogs by nature, their companies will be more agile if their boards are made up of a diverse group of foxes.
Agility means getting the most out of the time we put into projects. We can learn a lot from the agile software movement's use of pair programming. In pair programming, one person “drives” the keyboard and mouse, while the other watches, and they talk through the coding process. Pairs mix and match throughout the week to tackle different tasks. While this sounds wasteful, it has proven to be remarkably effective at reducing errors, cycle times, and support time. It keeps people focused and more engaged, helps eliminate bottlenecks, and gives teams more resilience. Pairs are supervised by a “scrum manager,” who keeps teams on track during the day and week. We could apply the pair-and-scrum-manager approach to critical business functions and measure whether it produces better results.
Diversity breeds resilience into an organization’s culture. It helps teams overcome obstacles and emergencies. Our aim is to help companies build cultures around resilience as much as capabilities. To do that, we:
Separate creativity and brainstorming from evaluation and decision
Eliminate blame
Eliminate performance pay
Implement evidence-based HR practices
Reward failure and integrate learning
Systematically eliminate groupthink
Implement feedback loops to help teams realize what actually worked, what didn’t, and what was due to chance
Match hedgehog CEOs with foxy diverse boards
Do boards of directors work?
Boardmembers can help raise money and find customers. But beyond that, they rarely provide the guidance or insights needed to really balance management and make robust decisions.
There are two important things going on here:
Groupthink. I’ve never been in a board meeting where someone says “I’m not so sure we should all be saying the same thing here, I think we haven’t really looked at the alternatives.” Decisions seem clear at the time, but later, when things don’t work out, you can’t look back at the minutes and see any warning signs. Most CEOs think of board meetings as a regular exercise they have to survive, rather than an opportunity to think the big issues through and ask hard questions.
Anecdotal reasoning. Most board members have had some success in business, and they believe this is all because they are so skillful. Like a doctor, they tend to remember a previous situation that was similar and suggest whatever worked in that case. There’s a huge difference. First, most doctors have already seen thousands of cases, not dozens. Second, most doctors aren’t very good at diagnosis and treatment, either, because they aren’t evidence based and they get poor feedback. In business, it’s worse.
Joan Garry has an excellent piece called 22 Signs Your Board Meetings Suck.