How Many Actions You Need For Conversion Testing

If you heard Trevo Claiborne from the Google Website Optimizer team speaking at The System seminar you already know his recommendation, to have at least ONE HUNDRED conversions/action PER OPTION, for each element you are testing.

Most standard rules of thumb circling around the direct marketing world say having at least 30 or maximum 40 conversion actions.

The reliability of rules of thumb always has and will bother one who has a statistical knowledge or experience.

It’s always best to have the appropriate statistical test applied, to get well based results. However, sometimes the practical need to move at the speed of business and use heurisitics and shortcuts could be considered as well.

Hearing Trevor say ONE HUNDRED in contrast to the 30 was incredibly surprising… This radical statement has the implications for us all but it’s reasonable to assume that Google knows better then anyone else…

Said that they must SEE something we can’t and my guess is that there is MUCH MORE NOISE on the internet compared to traditional direct marketing environments.

Now it makes clear, why so many people are confused with the results and do not see their profits really rising correspondingly, after having declared a winner.

As a result of that, you and I should all lean in the direction of longer lasting, more statistically robust tests that are based on the results of MORE actions before we make an educated decision.

The only problem is that many web site owners do not have the necessary traffic to accomplish the required tests in any reasonable time frame.

So what can be done? What is the solution?

You have to invest more time to plan better your split testing campaign and then perform FEWER AND MORE WELL CHOSEN TESTING OPTIONS that RUN LONGER.

And it’s quite reasonable because it really doesn’t make sense to test everything possible. If you carry out your research first and foremost, and determine what MIGHT be likely have the most impact on your conversions, you’ll do better.

How it can be done? You can survey people as they’re exiting your page… Live chats and monitoring competitive sites, reading repeatedly appearing sales copy in your market, can help you to identify favorite candidates to enhance conversions on your site.

• You can try and get more valuable results with a plain old A-B test with well thought out inputs, than testing umpteen different options.
• In such case, Taguchi (a more advanced method compressing dozens of tests into one) starts to look more appealing option, but you have to comply with the same rules… MORE actions than we’re used are required, to consider, before we get really robust and stable results.

And that means your inputs for Taguchi testing are even more important.

Bottom line?

Google says, since the results are not as “disposible” as you might have previously thought, and because of the LOTS AND LOTS of noise online, you have to plan your tests more carefully…
You have to perform longer tests, learn and better analyze the results to make sure you made right decision for long time.

Click here to learn more about How Many Actions for Conversion Testing »