Tuesday, 7 April 2009

Google does, apparently, test everything

I was recently interested by a debate about why Google sticks its facets, and now its query refinements etc, at the bottom of the search results. The basic assumption that was proposed was that you only need to refine your results if you didnt get it in the first 10 results, which you probably did anyway right.

I thoroughly enjoyed a good chat with Daniel Russell today, about this decision. I can reveal that it is a very very very well tested decision. Not just a random design decision, as I perhaps naively assumed. Apparently, they even tested 5px variations of it on the x and y axis, as well as placing it above and below the first result of 10 and many more options combined. And their high volume studies decided right there, not 5px to the side, was best.

Apparently it doesnt stop there. Even the height, and shade of blue, of the horizontal bar above the results has a dramatic effect. The colour blue has been carefully chosen.

In some respects, I feel like my research ideas and focus have just been completely shattered into tiny shards. But I guess I am now all the better for knowing (or believe I know) how purposefully Google is how it is. And its just like Daniel Russell said in the recent IEEE special issue, there are some things that you can only study at their level, including tiny UI changes.

Not that they only test small changes it seems.


Daniel Tunkelang said...

Dan Russell is no slouch, and I hesitate to second-guess his decisions. Still, I'd love to see something more substantive about what they studied, and what they learned. Should I assume from your minimal disclosure that you know more but aren't at liberty to share it publicly?

Max L. Wilson said...

I didnt get much more, and I'm sure Dan Russell was careful about what he disclosed. Still, I don't want to broadcast freely what he may have said to a small known group.

It would be nice, though, to see more. I can tell you, though, that the measure appears to be simply increased clickthrough.

Being right there gets the most clickthrough. Whether they care about anything else is questionnable ;) Despite the fact that Dan is running a workshop on sensemaking.

Daniel Tunkelang said...

Well, I am probably the last person with whom he'd share sensitive information about Google Product Search. Still, I'm surprised that he thinks Google is making a better design decision on where to place facets than Amazon, eBay, and Yelp, all of whom show them more prominently. Not that popular decisions are always the right ones, but those are popular decisions by leaders who do a lot of testing themselves.

Max L. Wilson said...

I think also that the context is different for Google. They are not a single retail point, and that probably affects it. So their answer is not necessarily right for everyone.

but they were purposeful in their design, which i wasnt sure about before.