Posted At : February 18, 2008 3:16 PM
| Posted By : Darcy Foster
Related Categories:
Web Analytics
- Don't pick the wrong page to test
- pick a relatively high traffic page
- pick a page near the top of your funnel process (from introduction to sale, the top is near the beginning)
- pick a page that has action - or calls to action
- Don't make the conversion page (test goal page) more than 1 click away from the test page
- keep the process simple (how do I get more visitors from here to there)
- Don't lose sight of test simplicity
- Even if you get hundreds of thousands of visitors - don't get all excited and start to test too many things on too many pages. Use a simple test strategy where the winning recipe can be rolled out to the rest of the site relatively easily.
- Don't make small vanilla changes that nobody will notice
- If you can't see the difference between version A/B/C from 3 feet away - how do you expect your visitors to? If the changes aren't big, the results will be too close, and the test will probably fail (no winner will ever be declared).
- Don't stop the test too early
- Relax, wait, watch and learn. If you stop the test before because one version looks like it is kicking butt right out of the gate, you could be making a big mistake by not looking at the big picture. Don't forget day of the week, time of the month, differences in where the traffic is coming from, etc etc. Ensure your tests run long enough so your data stabilizes.
If you like this post then please consider subscribing to our RSS feed.
I wanted to add a comment to your 2nd big mistake.
When selecting the conversion page for your Website Optimizer test it is always best to select the page that represents your ultimate goal (a lead being generated, a transaction being completed...). This page will often be more than one click away from the test page. It is important (when possible) to set up your conversion test so that it measures the change in performance of your ultimate goal, not just the change in performance of people getting to the next step. It is possible that you can drive more visitors to the next step in the funnel, but actual have fewer successful completions. This would happen when your test page starts speaking more compellingly to visitors who aren't actually going to convert, and less compellingly to those that would convert.
This being said many sites may not be able to select there end goal as their conversion page as they don't get enough traffic to their conversion page (too few actual conversions). If this is the case then it will take too long for your conversion tests to complete, and you should move the conversion page closer to the test page. When selecting the conversion page remember the following rule of thumb- each test should generate at least 100 conversion events per combination in a 30 day period. If you won't be able to generate 200 conversions for an A/B test in a 30 day period then you will need to move the conversion event up the funnel and closer to the test page.