Lesson: Building a Unicorn with Matt Mullenweg

Step #2 Talent: To attract the best talent, create a place where people really want to work

The entire Automattic.com site is somewhat of a funnel, and so we really started to think of the company as a product. And just like with any product you say, "What distinguishes us from everyone else? What is different about us? What would make you want to buy us," which is when you spend a third of your life working on our things that we work on.

One of those is that we do a lot with a few people. There's an outsize impact, certainly, per person, definitely per engineer at Automattic, and we think that can continue. I mean, we're growing very aggressively. We had 101 people last year. But still when you think about it you can, as an engineer, you can write code, deploy it yourself minutes later, and it can touch some of the largest websites in the world from CNN, to The New York Times, to all of WordPress.com which is a top 10 U.S. property. It's highly unusual.

Usually by the time that websites and products get to that scale, they become a little locked down for good reasons. Let's say, you're a bank or an oil company making tens of millions of dollars a minute, any amount of down time is probably unacceptable. But we try to have a Facebook-like philosophy of move fast and break things. I would rather make errors of commission rather than omission. I would rather us make mistakes because we were trying to do something in a better, faster way than the hidden opportunity cost of paucity of ambition.

We do have a fairly unique approach to hiring. The first thing that is a precondition for everything we do is to have some place that people really want to work. Without that I don't think most people go through hiring process because it's definitely little bit more in-depth than most companies. So first and foremost, we try to have publically a mission, something that people can understand and be attracted to or not attracted to before they ever even consider applying.

In terms of how we take applications, we try to do it in a fairly freeform way. We list a few positions on the website, which were permanently open. Always a more engineers would be great or more designers. And we ask people to send an email.

We have some little Easter eggs, so for example hidden throughout our site there are sometimes hidden code that will say, “Oh, you found this. Visit our job site and check it out,” or if we find that you visit the job site multiple times, I think five times, it gives you a little message that says, “You keep coming back. You might as well just do something. Pull the trigger.” Little things like that have been fun to develop organically over the years.

And then what we go towards is the entire automattic.com site is basically just a funnel for getting people to apply. It's really the only purpose that we even have a corporate website.

Once they apply, I do the first review of their resumes. The ones that pass that go to specialized hirers for the role. If someone specializes in hiring engineers or designers or finance or something, they go to a second pass and for the folks who make it past that, they do an interview on Skype text chat, pass that. If it’s an engineer, they do short code test, and pass that, and then we do the trial.

The idea of the trial is to create something that is as much like the actual work that they’ll be doing in their job as possible. Because the problem with interviews, this is a very artificial construct. You have something where it’s two people, high pressure situation for one, and they are talking, and you may be answering questions or things. It actually has a low correlation with what most people’s jobs are, certainly most engineers.

They are going to be in front of a computer and typing and thinking about problems and architecting things, and so who wanted the bulk of the time spent to be things that look like that.

Trials last anywhere from 2-6 weeks; the shorter the better. We’re not trying to have them be a certain amount of time. Basically what we look for is can we do a unit of work that allows us to really mutually evaluate for the person to see is this weird distributor company someplace they want to be and for us to see is this one of the best people that we can possibly work with. Is this someone we’ll be proud to call a colleague? As soon as we figure that out, we make an offer. Actually, what happens is they send them back to me. I kind of book end the process. So I do a final-final chat with the person, mostly looking for culture things. It’s very rare for someone to wash out at that level. It’s less than 1% or 2% and at the end of it I make a final offer to join the company.

Is the thing that you're measuring, the thing that is most important to the work that's been done? The thing that is touching the customer or the user. So in work when we are co-located in the same place you can't help but notice who's there in the morning, who's there when you leave at night, who appears sleepy and disheveled, who takes a nap during the day, who's you know on their phone all the time and doesn't appear to be working. And these things can influence your perception of someone’s’ work and productivity.

But there could be in early in the morning, leaving last at night, super well put together, well-spoken person who isn't producing that much and that disheveled homeless looking person that's producing some of the best stuff on the team. It's somewhat for free that when you're distributed you don't really know when people are napping during the day or walking their dog or whatever it might be. All you see is what’s emitted, so the product of whatever work they're doing.

If someone could work 30 minutes a day and have the same output as all the rest of our folks who work eight hours or 10 hours, good for them. That's fair because it's about the output and the expectations are fair across different, we call them Automatticians, people who work at Automattic. So by focusing on that, it allows people, I think, to change your life in different ways.

Obviously, people stop commuting when they work for us or they commute only by choice to maybe a co-working space. What they wear isn't as important as epitomized by the book written about Automattic called “The Year Without Pants.” It's a little bit of a joke. We don't usually wear pants. And then also how they divide their day. I work with folks who tell me, “I'm really most effective at night.” So I'm going to sleep 10 or 11 and really kick in the afternoon and they might go off for a few hours, and then be back kind of 10 to 1 a.m.

People have different cycles. There are medical conditions for people don't have 24 hour cycle. They have like a 25 or 26 hour cycle that kind of shifts throughout. I mean those things don't really matter to us. But they're very important when you're more of the factory mode of work where people are only working if they're in the office and doing whatever it is you think they're doing or pretending to.

I think that judging by output can cover anything. There are some positions where a time you’re there is important, for example, for the systems folks. We need to have someone available 24 hours a day. Maybe they don't have to be at a desk but they need to be near one. For support, we provide 24 hours support. So someone has to be working 24 hours a day. So then you have light weight schedule tracking but it's more peer to peer. It's not us saying you must work these hours or these hours. It’s everyone getting together and saying, “This is what my schedule is,” and someone else saying “Oh, we have a gap here, a gap here, I'll fill that in.”