The Oban Blog

Cultural multivariate can be testing for businesses!

multivariate mazeTypical multivariate testing procedure

For any company interested in multivariate (global or local) testing, implementing experiments can be a challenge. Typically companies new to the idea tend to run A/B split testing to answer questions from their marketing team: “Is it worth using Google checkout?”. The in-house web development team or external agency will then set up the pages to answer this question: “Yes definitely”.

More savvy companies might offload a lot of the setup and use Google’s Website Optimizer, great, but lacking in some respects, or another hosted services. These look to address the problems with Google’s multivariate tool especially with combinatorial explosion (see comment #1), and therefore the potential gains can be well worth the outlay.

But what about global sites?

For companies going or who have gone global, however, there seems to be a definite shortage of help available. Although you can buy any one multivariate optimisation tool, being careful of course that it can handle multiple languages there is a lack of consultancy for cultural multivariate, the knowledge of what to test to what culture. If testing fails at the first stage, the setup, there is no hope for conversion rate increase. Even if an experiment is put in place how do businesses know if they are testing something significant? When it comes to true culturally focused multivariate even the big players seem to be out of their depth, lacking the cultural knowledge to provide a worthwhile consulting service.

There are international SEO companies out there, but it seems to me that no-one yet has captured the cultural multivariate market.