What Marissa Mayer and Google Knows: It’s Speed

images1.jpgMarissa Mayer from Google was up next at the Web 2.0 Summit to talk about a key point Google has learned.

She started off telling us about an internal email list called Googlers, and they discuss things like the latest things in the snack kitchen, and if anyone can get Tivo Series 3 to work with Comcast.

In 1999 there was a ton of discussion about the Google Results page. What color the links should be, how many results should there be, and other things.

She was picked because she had no design experience to improve the results page

She set up A/B tests on the results page.

The first test showed that the more results they added to the results page, the less searches those people would do. There was a 20% drop in searches when they showed more than 10 results. Since their revenue has a strong relationship to traffic and search queries, this was bad.

The first theory was the paradox of choice. That users were getting frozen by too much choice. They found out that people who would get 10 results would get their results in 0.4 seconds. The experiment of more results took 0.9 seconds to get their results.

If you tracked how fast a search took, it hit 300 to 700 machines before getting back to the user. So Google learned very quickly that speed was important, and this is one of the reasons why they’ve done so much more in AJAX.

One of the product managers came to her to talk about Gmail, and said they wanted to do it in Javascript. Marissa thought they were crazy because it was buggy, slow, client side, etc. The product manager said she was right about all of it except it being slow. They moved on to really focus on speed with Gmail.

They did a similar thing with Google Maps. It was bloated at first, and they put it on a code diet and managed to trim it down and speed it up. They saw traffic results rise in direct correlation.

She pointed out that instant feedback leads to a steeper learning curve. She pointed to digital photography as an example. People are becoming better photographers because they can get feedback on their photos so much faster. They can look right after they take them, they can post them to the web quickly and get feedback from others, and improve their work. In the old days, it took much longer to see your pictures, so photographers improved slower.

This theory holds true for Adwords as well as users could get their ads approved and running right away, which was a key reason they gained market share vs. Overture which had to approve ads that often took days.

A negative example for Google was Google Video. Which initially took 24-48 hours after you uploaded a video for it to show up. YouTube on the other hand posted it instantly. Which one rocketed in growth? YouTube!

She pointed out the mobile space is suffering a bit because applications on mobile are still too slow. After we see more advanced in mobile hardware we’ll get more usage.

It was a really fascinating 10-minute talk from Mayer. Man she’s efficient!

  • http://www.thoughtshapers.com Jeff Molander

    “That users were getting frozen by too much choice.”

    Hold on there! Users might also be simply finding what they want when they searched less. More results on a page increases the likelihood that someone will not need a second page, right??

    GOOG has certainly addressed the speed issue since 1999 and can serve 10+ faster than .9 seconds in most cases. I’m left feeling like they did this to increase quantity of searches although this doesn’t really indicate opportunity for more search ad inventory, just more search queries.

  • http://www.conversionrater.com/ Pat McCarthy

    Hi Jeff,

    She said when they tested moving from 10 results per page to 30 results per page, total traffic and search queries dropped 20%. Total traffic dropping would have a negative effect on ad inventory.

  • Pingback: Optimize your pages – Part 1 « Webs Blog

  • Pingback: Optimize your pages – Part 1

Post Navigation