Shared Command

Published on July 2, 2013
by Ben Horowitz

Cause two suckers can’t agree on something, a thousand mutherfuckers die for nothing.

— Geto Boys

Battle is a highly fluid situation. You plan on your contingencies, and I have. You keep your initiative, and I will. One thing you don’t do is share command. It’s never a good idea.

— Major Vic Deakins, “Broken Arrow” (played by John Travolta)

Zynga recently put a new management structure in place where Don Mattrick is the CEO and my friend Mark Pincus, the founder and former CEO, will be the chairman of the board and the chief product officer.


My friend Fred Wilson encourages entrepreneurs to consider this structure and wrote that the new Zynga organization is analogous to LinkedIn’s, with founder Reid Hoffman as chairman and Jeff Weiner as CEO. I think Fred is wrong. The Zynga structure is quite different. At LinkedIn, the chain of command is clear: Jeff Weiner runs the company.

If you want a decision made and Jeff weighs in, that’s it. Game over. Reid never overrides Jeff. Contrast this with Zynga, where Mark reports to Don who reports to the board where Mark is chairman. Who makes the final decision on products? If a product decision impacts revenue, who makes that decision? I do not know the details so I do not know for certain, but the Zynga structure smells much more like shared command.

If it is indeed shared command, that will be a big problem.

Shared command always seems really attractive to the people at the top of the organization like the CEO and the board: “We have two world-class people, this gives us the best of both worlds! We shouldn’t get caught up in the conventions of years past. We’re all adults. We can get along.” It looks much less attractive to those who do all the work in the organization. To them it looks more like frustration, chaos and delay.

As a company gets big, the information that informs decision-making gets massive. Depending upon the prism through which you view the business, your perspective will vary. If two people are in charge, this variance will cause conflict and delay. Every employee in a company depends on the CEO to make fast, high-quality decisions. Often any decision, even the wrong decision, is better than no decision. These decisions are the pulse of the organization. Sharing command almost guarantees that the CEO position will perform poorly in this dimension.

Let’s look at the problem from the point of view of the employee. Imagine you are an engineer and you need to know if you should optimize what you are doing for IOS or Android. Both platforms are “first class” as far as the company is concerned, but you are going to have to make a trade-off for this particular function. So, you run the decision up the chain and it gets to the CEO. Excuse me. It gets to the chief product officer/chairman. Excuse me, which one must it get to? It impacts revenue, but it’s a product decision. Or is it? Whomever she checks with will have to check with the other, who may be in China and unreachable at the moment. Maybe the CPO/chairman and the CEO are at odds and need to wait until they are face to face to resolve it. It the meanwhile, she waits and waits.

Now multiply this by thousands of employees and that’s what it means when you share command.

If shared command is so bad, why do people keep trying it? Basically, the people in charge need to make one decision: Who should run the company. They cannot decide, so they compromise. And now, because they couldn’t make one decision, every single decision that the company makes from this day forward will be slower than it would have been otherwise. It sucks to work for that place.

The Great Exception

You may be reading this and asking, what about Workday? Aren’t Dave Duffield and Aneel Bhusri co-CEOs? Isn’t Workday the greatest new enterprise software company known to man? The answer to all of those questions is yes. Workday is the great exception to the rule. Why is Workday able to do it? What makes them the exception that proves the rule? In speaking with Aneel, several factors make Workday’s command structure unique.

  • Aneel and Dave have been working together for over 20 years — they are not just co-workers or friends; they are family. They comprehensively know and agree upon each other’s strengths and weaknesses. The have total trust.
  • Dave is 26 years older than Aneel — there is no chance for a sibling rivalry. Dave is far more excited about Aneel’s success than his own and vice-versa. Dave sees Workday as Aneel’s company eventually and is there to help Aneel succeed. To a large extent it is Aneel’s company now in terms of command structure.
  • Command is not actually shared at Workday, command is partitioned. Because Dave and Aneel know each other so well, the division of decision making is natural and obvious to them and everyone in the company. Beyond that, since Aneel is the eventual lone CEO, the decisions are his to make and get Dave’s advice when he needs it. Aneel gets the bonus of being able to completely delegate whole categories of decision-making and responsibility to Dave whenever he needs to.

So, if you’ve worked together for 20 years, have already selected who will eventually run the company and have a track record of being two of the greatest entrepreneurs of the past two decades, then go ahead and share command. Otherwise, it is never a good idea.

Footnote: Some people perceive that Marc and I share command at Andreessen Horowitz. This is not true. The operational reporting hierarchy, where most decisions get made, is run by Scott Kupor. The investment decisions are decided (as in other venture capital firms) by the general partners.

Ben Horowitz is founder and general partner of Andreessen Horowitz. He was a co-founder and CEO of Opsware (formerly Loudcloud), which was acquired by HP, and ran several product divisions at Netscape. He serves on the board of companies such as Foursquare, Jawbone, Lytro, Magnet, Nicira and Tidemark, and blogs at www.bhorowitz.com.

Return to: Shared Command

URL: http://allthingsd.com/20130702/shared-command/