03 June 2006

Taguchi again

The last post about multivariate analysis might have seemed a bit daunting - it's really just a way of looking at a variety of possible EITHER A OR B options in parallel. The point I should have made is that Taguchi analysis lends itself to relatively high volume sites where developing the data from the test groups doesn't take too long.

For sites with different visit characteristics then you have two options:
  • back the experience and judgement of the people who have built the site
  • do focus group testing with a couple of people who haven't been deeply involved in the design and execution of the site
I would always recommend testing, irrespective of how much you trust the judgement and skills of the team who designed and built the site. Remember, what you are interested in here is not statistical validity - merely a commercial truth. Does the headline work? Is the offer clear? Is the offer compelling?


Post a Comment

Links to this post:

Create a Link

<< Home