Harper Reed: how to build teams, practice failure and learn to listen.

The man who was CTO for Obama's 2012 election campaign brings us the most important lessons from their 18 months building victory.

Harper’s first computer was an Apple one. He was six, and started programming with his brother. He the read Hackers: Heroes of the Computer Revolution. He loved their ethos. He wanted to pick the lock and see what was behind it, just like hackers did. He wanted to be a hacker. He thinks “hacker” and “coder” can be used interchangeably.

Then he joined Threadless – they invented crowdsourcing, and didn’t realise they had done it. They just wanted to make t-shirts. Someone designs a shirt, people vote on it, and cash falls from the sky. They have over 100,000 designs – and grew from single figure millions to double figure millions in revenue.

He realised he’d achieved his goals – and so he quit and went on a vision quest. Which means he wandered around and worked in venture capital for a while. He was going to start a startup – but then the Obama campaign came calling looking for a CTO. The real question is: Why Harper? The question doesn’t come up so much since they won, but before it came up a lot. Well, if you want rice cakes, you go to the rice cake dealer…

In politics that seems insane. They don’t hire engineers. The Obama campaign hired 40 engineers. They had 120 tech staff. In 2008 they had 4 engineers, and between 12 and 20 technical staff. And they had 18 months, which was longer than their opponents. And they knew that they would raise about $1bn. He focused on the one thing he knew how to do: execution. They built a platform. It was called Narwhal. There was a lot of trust issues between the hackers and the rest of the team – so they tried their best to bridge the gap. They photoshopped the campaign manager’s head onto various animals…

But Narwhal was a concept – how do they execute fast? They needed an API. That would allow them to build a foundation, which would allow them to build products like dashboards, social apps and so on. They had 200 or so of them… They focused on what solved the problems they had. “The fate of the world was in our hands”. They focused on user-experience. They practiced failure for a month – they tested their failure plans. By the time they got to election day, they were calm. No changes once they hit election day – so they just chilled out. They won.

But politics is hard. Larry Macow told him to manage by his outbox, not his inbox. Problems with your co-workers are probably due to communication confusion – if you reach out, they might embarrass you with their response. People can feel under-served by you.

Building Teams

Building teams is hard. Pruning is the first step – and it’s hard, because it means firing people. He fired someone from Threadless that he then hired for the campaign. All it meant was that he fitted in one role and not another.

ABC – always be creative – you need to pitch your company well to get good people. You need to say “we’re solving the world’s problems – and the problem is x”.

As hire As and Bs hire Cs. So hire As. Trust is important. Measure everything – not enough people in the audience are pro-metrics. Diversity is important – don’t be afraid to hire people who look different from you. He failed – everyone he hired looked roughly like him. This told him that’s it’s hard to find women who want to work in crazy tech jobs who have 8 to 10 years experience. Try. Talk about it when you try. The internet is about democratisation of access – not everyone is a white dude from Chicago with a  red beard. It will not get easier until we all try. Practice success. Celebrate success. They gave the team credit. People got to see what everyone else was working on.

But the most important thing is shipping. His challenge to the audience: if they have an unshipped product, ship it by Monday. It’ll be fine. He promises.

Practice Failure

Failing sucks. Work hard at being terrible at it. The first way they dealt with it was user experience. If you engage and invest from the start you don’t have the silly conversation about what the users like when it comes to testing. It’s the difference between functional and usable. They did loads of A/B testing. Groundhog Dayis just a  movie about multi-variant testing. (There are blogs that estimate that the main character spent 500 years in that world). The e-mail team would bet on subject lines and which one would get the best response – no-one ever won. They couldn’t guess, they had to test. Never roll back code – only roll forwards. It’s much happier. The shuttle programme had the idea of fail safety – to make sure the people in the shuttle came home. Why don’t we do that for users? The users should never see an error as an error – it should push them to somewhere else in the site where they can contribute.

They practice through Gamedays – they constantly destroyed everything. They played the game of “what if”, figuring out what could go wrong, and building a process for it. They had no downtime thanks to this.


The media have no idea why they’re talking about when it comes to big data. Micro-targetting is clearly really exciting…They used e-mail data and Facebook and texts. He was worried about donations by text – but actually it worked well, and raised boatloads of dollar bills, and allowed better content distribution. They could get friends to encourage friends to vote. It got action – and no-one said it was creepy, apart from the press.

Awesome, right?

Well… we’re often bad listeners. Tim O’Reilly said “you should stop microtargetting, and focus on micro listening”. You need to have conversation. How do you make those strong ties? They spent a lot of time making sure door knocks lead to valuable conversations. They sent DMs on twitter to people who followed the president – it blew their minds. As did DMs from the first lady. The recipients knew it wasn’t really them, but they were still excited. They targeted people with influence.

They asked a lot of simple questions. But they had to force themselves to listen. This, along with diversity, is the most important lesson.

Their software was open source – they built it with the ideals of the Democrats built into it. He thinks a Republican would be puzzled that they did this. But it’s no longer state of the art, and so will not immediately help the enemy…

Answers to questions:

He’s doing a startup now – built around the idea that a shopping cart is a ludicrous metaphor to be using on mobile.

The youngest team members was between 18 and 20 – and was frustrated by being told he couldn’t fraternise with the interns…

They used Ushahidi to great effect in places like Virginia.

He didn’t have to interact with the president to do his job, but did meet him a couple of times. He was less aloof and more normal than most executives, and really, really funny.