Tech company must better at 'dog-fooding', says search quality chief.
Dan Russell‘s official job title is uber tech lead for search quality and user happiness at Google. He teaches people how to use Google – sometimes with an attentive Sergey Brin lurking alongside.
He is one of the battalions of people at Google tasked with improving search, but was also part of the small team who famously tested 50 different shades of blue on Google links to see which got more clicks. Data is king at Mountain View.
“We were playing around with colours, and one of the questions that came up was what if we removed the underlining – so we did that – and it turns out people don’t like it,” he said of the experiment 18 months ago. “It slowed everybody down by a significant amount.
“We got all this data and didn’t really understand it. Just that some shades did better than others. Mostly if you look at it you can’t tell, but behaviour changes.”
Russell joined Google in 2005, when the company would be running a couple of dozen experiments at any one time on how its users search. Now Google runs hundreds at the same time. (Russell repeatedly declined to show me examples of tests happening right now.)
“One thing to know is that Google is always experimenting, and every time you see it something is different. We’re doing millions of tests,” he said.
Although Russell hasn’t lost his bounce, this obsessive attention to detail grates on some. The designer Douglas Bowman left Google in March 2009 with a revealing parting-shot at his former employer, you’ll recall. Bowman, who has since joined Twitter as its creative director, said at the time:
“I had a recent debate over whether a border should be three, four or five pixels wide, and was asked to prove my case. I can’t operate in an environment like that. I’ve grown tired of debating such minuscule design decisions. There are more exciting design problems in this world to tackle.”
Russell is undeterred. What does Google need to get better at? “Dog fooding,” he says. “Sometimes the truth is we launch something [feature changes] when we think it’s good enough to get us to the next step, and we need to actually test it on live users rather than just Googlers. So that’s the dog-fooding step.”