Recently I had the opportunity to sit down with CTOs from very different digital media companies to understand how they think about the role, what they have in common and where they differ in approach. You can read earlier interviews with the CTOs of Vice, Salon, Shutterstock, Digg and Hearst here.
Next up is Mark Wilkie, CTO of BuzzFeed, one of the fastest growing media companies in the world. We discussed how BuzzFeed has evolved since its launch, and managing technology in a business that is going global.
What’s different about BuzzFeed in June 2014 versus the company you joined a few years ago?
I’ve been at BuzzFeed since the beginning. One of the things I love about working here is working with our CEO and Co-founder, Jonah Peretti. Jonah understands technology and he sets us up for success. Building software teams is difficult, taking abstract ideas and turning them into software is tough. Having someone on your side that trusts you to do the right thing is great. Jonah knows what we need to succeed.
These days, I see BuzzFeed as a technology driven company; the core of what we we are doing is technology driven. Technology is a first-class citizen; everyone sees that. The business and editorial teams understand that. I have a team of nearly a hundred across IT, engineering, product and operations. And, we’re going global. We are launching different verticals in different countries; we are trying to accent the team with international support when necessary.
We hear all the time that the pace of technological change has sped up. How has that changed the job of the CTO, particularly in digital media?
The thing that has sped up is the way people communicate around technology. When you think about the community around tech, there are more people doing it and talking about it. So things move faster because people hear about developments faster. There is more conversation- more meetups, more reporting- information gets out there much faster. The days of Linus Torvalds discussing how to improve Linux on a Usenet group are gone.
We live in an interesting era where open source is embraced in a big way. People are building amazing software leveraging great collaboration tools like GitHub and Stack Overflow. People are finding great tools and sharing them; that just encourages software to grow and get used more. The ecosystem is just a lot more vibrant. There used to be a question about whether we should use open source tools; these days no one questions using open source software.
How much tech do you buy vs build?
This is always the grand debate. For instance, look at content management systems. At the Huffington Post, they started with Movable Type. But if you looked at that code base today, you wouldn’t see Movable Type. When we started building BuzzFeed, we thought about that, and we decided to build what we needed specifically. It’s evolved and grown; that’s the nature of software. Building our own tech stack and being vertically integrated makes a lot of sense for us.
Of course we don’t want to build a CDN- we use Akamai for that. And, we leverage a lot of support services and tools. We could build our own push notification infrastructure. But we’re not going to do that because we can leverage a company that is already doing it. We can get a product out the door and working quickly and evaluate it. We do that a lot- use a third party tool to evaluate a concept before we decide what direction we want to take.
In the modern age, if you leverage a third party to do things and you don’t control it, you’re dependent. You are depending on the third party to be as good or better at infrastructure than you are. You have to trust those third parties to be robust. And you can’t know how reliable they are until you use the product. Then you are all in; and then it may be too late.
One of the big things we are looking at right now is how to take the data science architecture to the next level, how to build the tools that our data science team can get excited about. That’s an area where we are looking for outside expertise.
Getting things out, testing things, being as nimble as possible is the key. Build, test, integrate, deploy. We strive to get code out in the production ecosystem as soon as we can. We want to get feedback from users and our systems early in the process. Then we iterate, improve, and the cycle starts again.
What’s the secret sauce to BuzzFeed’s technology platform?
When we were building BuzzFeed we focused on the social world. How do we know when things are good, how can we give feedback to editorial; everything was focused on sharing and social. So yes, we were building a CMS. But at that time it was about aggregating content and data from other places, figuring out how people were using that content and sharing it with friends. We wanted to be able to see how things were accelerating, building data and analytics that helped us reduce the friction from sharing.
What types of time horizons are you working with when thinking of how to incorporate new technologies into the business?
This idea of thinking out five years, that seems crazy to me. I’m not thinking about technologies five years out. On the business side we think that far out; but right now in the tech team we are thinking about tomorrow. Of course, when I think about BuzzFeed as an organization in five years, I do think about the how the tech team will change as we get bigger and grow. How do we manage when we are twice the size? How do we maintain the level and quality and speed of software?
That leads us to talent. The talent game is tough. But as we get bigger it gets easier to some extent; it’s easier to attract people. The big thing is finding people who are excited about BuzzFeed, about who we are as a company. Technically proficient is one thing, but they can’t be blasé about BuzzFeed; we want them to be invested in who we are. Developers who know PERL, mobile, Python- we are hiring. The hardest thing to find is a maestro DBA. We are looking for great DBAs who fit the culture.
How do you personally keep up to date on the latest?
I don’t do as many Meetups as I used to, but I have good friends whose Twitter streams are interesting. We do these Tuesday tech talks where we have people present internally about something they are working on, or a technology they are excited about. It’s the way devs can pitch stuff to us, to advocate for an idea. We want everyone on the teams to be able to see what we’re working on. We invite them to do ten minute lightning talks. The whole team comes to those. We do three tech talks a month and then the second Tuesday we call it tech brews. We get pizza and beer and we get someone outside the tech team to present, perhaps from sales or editorial. They talk about their view of technology in the company and how it helps them, and what they need.
We do host meetups. The Django-NYC meetup, we love to do that; it’s easy, the devs like it, and it’s good to get people talking about specific technologies. It’s also a nice recruiting tool.
This is the biggest team you’ve run. How are you scaling up your own approach to the job?
I read and talk to people who have been there, like Paul Berry. There’s a lot to be learned from his experience working at Huffington Post and RebelMouse. Twitter’s former SVP of Engineering, Chris Fry, wrote a nice piece on scaling engineering teams a few months ago.
I feel like we’ve been really successful. Part of it is trying to be a bit organic about identifying what is working and part of it is realizing when you’ve made a mistake. Every year at BuzzFeed is like working at a different company. It’s a different job, it’s a bigger team, exploring different things and different technologies.
You’re seen as a disruptor right now. What could disrupt BuzzFeed?
There are a lot of BuzzFeed copycats now. For instance, we started working on quiz technology a long time ago; now there are a lot of quiz knockoffs. But we are not trying to make one-hit wonders; we are focused on building great technology that is valuable to us and to our readers. We have a great tech stack that helps us iterate and provide data to know what’s doing well and what’s not; we have great product people working with the editorial team. People can copy us, but it will be difficult to copy the model- such great teams working with great technology.
A big part of what we do is the combination between editorial instincts and technology. That’s one thing that’s always been amazing about Jonah. He’ll have an idea and we’ll say, that seems a little crazy. But we implement, test, and a lot of times he’s right. Of course, we have the data to back it up.