I was struck by how many Steve Jobs biographers have emphasized how important it was that Jobs got kicked out of Apple. That experience changed him and in the long-run, made him a much better businessman.
Pride before the fall is a truism and even being aware of it doesn’t prevent one from falling into into that trap.
Right now, as I write this, Stardock is doing better than at any time in its history. But it’s only doing so well now because of changes we made 3 years ago. Because 3 years ago, the company was in chaos…
A trip back to 2010
Our main business has been selling software and it had had a steady string of big successes. In 2010, our software came with virtually every Dell, HP, nVidia and ATI card and Parallels for the Mac was using our tech to make Windows apps look like Mac apps.
At the same time, Impulse was growing at a dramatic rate. On the one hand, it was only doing 1/10th of what Steam was doing, but on the other hand it was doing 1/10th! of what Steam was doing!
And the games side of the company? It had had the dual hits of Galactic Civilizations II and Sins of a Solar Empire.
It seemed like we could do no wrong.
Here was the company structure in 2010:
That sound is the collective gasp of business managers reading this saying “You. Guys. Were. Insane.”
As ridiculous as that structure looked, it was actually a lot worse than that because I also directly programmed on various projects as well and sometimes created artwork and web design for the websites.
The first signs of trouble came with a game called Demigod. While Demigod was a very good game, its launch exposed just how thin we had spread ourselves. On Good Friday in 2009, GameStop retail put the game on the shelves (it wasn’t supposed to come out until the following Tuesday). Stardock was closed for the Easter holiday and the game, lacking copy protection was soon out everywhere to the point that by Easter – 2 days before the game was supposed to ship – there were over 100,000 IPs trying to connect to he proxy servers per hour. There just weren’t enough people (at Stardock or GPG) familiar enough with the project to scale the network structure in place fast enough to address the problem quickly.
Simply put, if it takes N seconds to connect one player to another where N is the number of players trying to connect, we just didn’t have enough people to set up enough servers to do this quickly if you had that many people playing. We should have, btw. But because of the lack of depth in the management chain, there wasn’t anyone dedicated to doing the simple math. In other words, if one person made a mistake (such as not properly calculating how many servers were needed to handle say 10,000 simultaneous connections) there was no backup.
The way it should have been done is that there should have been a manager who oversaw a Stardock manager and a GPG manager who in turn had their own teams. Instead, there were no managers on either side. It was both teams chatting via Skype directly. Because of the sheer raw talent at GPG and Stardock it worked amazingly well (if they hadn’t miscalculated the # of proxy servers needed to make connections really fast no one would have noticed). In other words, as bad as those first couple weeks were, if the teams involved weren’t so talented, it would have been much worse.
Demigod was a warning sign. But it wasn’t recognized.
War of Magic
In Summer of 2010, the company’s management system had largely broken down. Each manager essentially had to operate on their own because the top level management (me) was spread across 11 direct reports. I was working around the clock by this point because not only did I run the company but I was programming on the game.
Stardock’s best game developers had been working on Impulse and the majority of those working on Elemental had never worked on a game before. In other words, unlike Demigod, my management failings wouldn’t be masked by having super talented veterans bailing us out.
The new developers received no mentorship, no real training. They had been thrown into an impossible situation. We fell into “It’s faster if I just do it myself than to train someone” to the point that 100% of the quests in War of Magic were personally coded by me via notepad and XML (very tedious).
Moreover, with so little management structure, decisions were made by managers with very little oversight (for example, the release date for Elemental was picked by the BizDev team supposedly because it was easier to get shelf space in August).
To be specific, if a company’s leadership is bad (like having a CEO that’s spread between a dozen groups and programming) then managers will fill in the void based on what they think is right or what they think is best for their group or themselves. And there was a lot of this going on. Decisions had to be made and if the CEO/President/Founder wasn’t around because he was messing with XML at 4am those decisions still had to be made by someone.
The result was our first truly serious failure in almost 20 years in business. The failure required us to look and see what happened. One thing we saw was that if the decision making tree is too broad it becomes very fragile.
Fast Forward
Stardock in 2013
Now, looking at the chart you might see it isn’t perfect. I could certainly draw up the ideal org chart. But the real world (available management talent) dictates a lot of this. Excellent managers are shockingly rare – I suspect many of you reading this can verify this from your own experience with managers. Going from 11 direct reports to 4 has ensured that the decision making process is a lot more thoughtful and not driven by fatigue or ability to put time into something.
There’s probably a book
Somewhere I am sure there are some great books that go over the maximum number of direct reports and the depth of management structure per 50 employees (feel free to recommend in the comments for all of us to check out).
The point is, if you have success after success it becomes very easy to resist necessary change. I hope my experience helps you in your own endeavors in the future.