15 October 2006

The cost of sale

We have a site which has been running since the middle of August. We have been driving traffic towards it using AdWords and the cost of the qualified clicks has dropped by about 12% over that time. I think that we'll be able to drive it down further by refining how the AdWords are used and having parallel campaigns which focus on particular keywords.

That's all good, but the cost of sales is still too high. I have been doing some research to find out some independent opinions of the site to discover why the site isn't converting a high enough proportion of its visitors. There are a number of areas that the participants think that we need to change and I will be sharing the outputs of that survey with Chris and Phil. We won't necessarily implement it all blindly, but we will try to think through what we might reasonably do to react to some of their comments. The point here is that the sample that we have used for the research isn't statistically significant, but there is still some commercial truth in what has emerged from the research.

Experimentation is key to any eCommerce site. There is absolutely no point in increasing the resources devoted to promotion or search engine optimisation unless you are confident that the traffic has a reasonable chance of converting. Higher AdWords traffic to a site that isn't selling is simply a strategy designed by a liquidation specialist looking to make money from another bankruptcy.

That means changing things and measuring people's reactions to those changes. What is the dwell time on the site - what is the average number of page reads - how many go through to the sign-up page - how many hand over their card details? Making changes slowly and then checking to see whether those changes have a beneficial impact on the figures takes time, but it is worthwhile when it pays back with a higher average sales value per visitor to the site. If you are lucky enough to have a high volume of visitors then it can be worthwhile using some form of Taguchi modelling, but if, like this site, the number of impressions is relatively low then you will have to make judgements based on simple individual research and focus groups to get an insight into whether the changes you are implementing on your site will be beneficial.

Labels: ,


Post a Comment

Links to this post:

Create a Link

<< Home