Welcome to WebmasterWorld Guest from 54.224.180.131

Forum Moderators: Robert Charlton & andy langton & goodroi

Message Too Old, No Replies

SEO aspects of A/B testing

     
3:00 pm on Jul 20, 2011 (gmt 0)

New User

5+ Year Member

joined:Dec 14, 2008
posts: 3
votes: 0


Hi,

Hypothetical scenario:

You want to try out the long term effects of a different layout, and decide to serve 1 out of 500 users this new layout. You remember the randomly assigned 'beta' visitors, so that you can keep serving them the new layout and track their long term behavior.

If one of Google's bots is one of the 'beta' visitors, and another one is not, they might interpret this as if you're trying to mislead them?

What is the safest way from an SEO perspective to do this?

Thank you in advance,

Raf
4:27 pm on July 20, 2011 (gmt 0)

Administrator from US 

WebmasterWorld Administrator goodroi is a WebmasterWorld Top Contributor of All Time 10+ Year Member Top Contributors Of The Month

joined:June 21, 2004
posts:3241
votes: 202


Can you better define long term? Are you talking a month, a year, forever?
4:50 pm on July 20, 2011 (gmt 0)

New User

5+ Year Member

joined:Dec 14, 2008
posts: 3
votes: 0


About 2/3 months
6:11 pm on July 20, 2011 (gmt 0)

Preferred Member

10+ Year Member

joined:Jan 5, 2003
posts: 389
votes: 0


Google offers the testing that you are talking about -- [url]www.google.com/intl/en/websiteoptimizer[/url] -- so I doubt that there would be SEO implications unless the method you use to do the testing is severely out of the ordinary (not to be read as, "unless you don't use Google").
7:51 pm on July 20, 2011 (gmt 0)

Junior Member

10+ Year Member

joined:Aug 31, 2006
posts:69
votes: 0


Most testing suites will handle bots so that they don't see variations. Unless your doing some sort of server-side segmentation you generally have nothing to worry about.

Serving 1 in 500 visitors may require quite serious traffic in order to return significant results.