A/B Testing + Duplicate Content Paranoia

While reading around on the Google Website Optimizer forums this morning I noticed several concerns regarding “duplicate content issues” while A/B split testing with Google Website Optimizer. Some of the time we are strictly testing “layout” options, so we end up with test pages having the same content as the original. This isn’t something I’ve ever considered an “issue” before, but since it’s such a popular question I thought I’d share my input on it.

First off, I don’t think duplicate content penalties are something to live in fear of. It’s not even really a penalty, it’s just Google’s idea of which URL is “best”. For a better understanding of how Google handles duplicate content go here.

Second of all, I don’t usually see variation pages getting indexed as there are generally no links to them since they are redirected from the original page.

If this isn’t enough to put your mind at ease, continue reading …

Google is very good at picking the best page to display in results, but I’ve included some things you can do to make sure that your “preferred version” remains the same while testing. I would probably only go as far as step 3 in most cases, but have included every precaution to be taken for the truly paranoid tester.

1. Include your original page in your sitemaps (HTML & XML). You should probably be doing this already.
2. 301 your variation pages to the original page after you’re finished testing. Depending on the length of your experiment your variation pages could possibly build up some link juice. This will ensure that any link power is redirected back to the original page.
3. Specify your preferred version with the new canonical tag. You can place this tag on your variation page(s) to let the search engines know which is the preferred version of the page.
4. Exclude your variation pages in your “robots.txt”.
5. Add noindex tags to your test pages.

That concludes my list of precautions you can take against duplicate content while testing, but for the record I wouldn’t lose much sleep over this subject to begin with.