The Billion Dollar HTML Tag

23 comments

Marissa Mayer

Marissa Mayer of Google discusses web site performance Wednesday at the Velocity 2009 conference (photo by James Duncan Davidson via Flickr).

Can a single HTML tag really make a difference on a corporation’s financial results? It can at Google, according to Marissa Mayer, who says web page loading speed translates directly to the bottom line.

“It’s clear that latency really does matter to users,” said Mayer, the VP of Search and User Experience at Google and today’s keynote speaker at the O’Reilly Velocity Conference. Google found that delays of fractions of a second consistently caused users to search less. As a result, its engineers consistently refine page code to capture split-second improvements inĀ load time.

This phenomenon is best illustrated by a single design tweak to the Google search results page in 2000 that Mayer calls “The Billion Dollar HTML Tag.” Google founders Sergey Brin and Larry Page asked Mayer to assess the impact of adding a column of text ads in the right-hand column of the results page. Could this design, which at the time required an HTML table, be implemented without the slower page load time often associated with tables?

Mayer consulted the W3C HTML specs and found a tag (the “align=right” table attribute) that would allow the right-hand table to load before the search results, adding a revenue stream that has been critical to Google’s financial success.

Google tests the relationship between page load times and usage in regular experiments with subsets of users. “Injecting delays consistently produced a drop in searches,” said Mayer. An example: the number of search results, which defaults to 10 results. When some users wanted to see more results, Google tested pages offering 20 or 30 results, which added between 0.4 and 0.9 seconds to the load time.

The outcome: users seeing the page with 30 results would up conducting 25 percent fewer searches, a trend which could equate to billions of dollars of lost revenue. Google allows users to customize their search settings, but the default remains set at 10 results.

The oddest HTML hack involved the Google Checkout “shopping cart” icon that appears next to ads for vendors using Checkout. Pages where the icon appeared were loading slower than other pages, “It’s such a little image,” said Mayer. “It shouldn’t make such a big difference. You actually have an image opening another connection, and connections are expensive.”

Google engineers found a kludgy but utilitarian solution in HTML table code. “They actually found a way to draw the Google checkout cart using HTML,” said Mayer. “It renders faster. It’s interesting to think that HTML and images can affect user impressions of speed.”

About the Author

Rich Miller is the founder and editor at large of Data Center Knowledge, and has been reporting on the data center sector since 2000. He has tracked the growing impact of high-density computing on the power and cooling of data centers, and the resulting push for improved energy efficiency in these facilities.

Add Your Comments

  • (will not be published)

23 Comments

  1. Marcus

    Correlation does not equal causation. Have they considered that a result page of 30 results instead of 10 may give answers with less searches? Other explanations?

  2. Wow. That... really sad considering how much slower other search engines were compared to google. I mean, if they conducted 25% fewer searches, did they just decide that finding their thing wasn't worth it? I think it also describes a sad lack of determination in folk in our society. If you're page takes .4 seconds longer to load and that's just too much for you to take, I mean, how could you possibly even pay attention to a short-attention-span theatre like Robot Chicken?

  3. @Marcus - Excellent point. It would also be interesting to know how many users clicked through to the 2nd or 3rd page of results to find their answers as this behavior would also result in less searches. But the article raises some very valid points, particularly about the cost of connections. I'm still amazed how many big company websites still fail to address these issues when they have plenty of resources to do so.

  4. In the Q&A Mayer was asked whether the dropoff in usage was because users found the best result on their first screen and thus didn't need to look for results 31-60. She said this occurred to them and they dug deeper and found that the subset of users seeing 30 results were not just seeing fewer pages - they were actually returning to Google less often. I think this occurred to lots of folks. After all, for 30 results Google displays three pages of ads with a limit of 10, and just one page of ads with a limit of 30 results. Nonetheless, Mayer said the longer page load was apparentlyy causing some users to go elsewhere.

  5. Jack

    @Marcus I think Google's ability to do statistics and to analyze issues related to search are quite solid. Particularly in this case, the ability to do experiments with their search homepage over and over again billions of times over, would give a very strong counterfactual dependence for causality. This is their bread and butter afterall!

  6. cak

    No marcus, you are the first one in the world to study first year statistics, nobody else, on the entire planet has ever thought of tailoring their experiment to rule out causation. Thanks for sharing your deep thoughts with us... what is next you are going to solve middle east crisis by suggesting they should all just be friends? I can't wait.

  7. Not A Tag

    An HTML attribute is *not* a tag. It is an attribute. I find it amusing that Mayer, who has a reputation for being somewhat anal, and insisting that things be done "right" (in other words, her way), would entitle an article so "incorrectly". Of course, I guess "The Billion Dollar HTML Attribute" doesn't have that nice buzzword sound to it.

  8. This is a very interesting post. This brings to the fore the need for testing, testing, testing everything no matter how small or insignificant they seem. At times what looks most insignificant might end up being the most important!

  9. load times have always been important. important for seo, bounce rate... great info btw