Stats usage

If your trying to improve the conversion rate yet you don’t have that much traffic going through the page, which is very usually you need to approach it in a very specific manner.

Limit your options.
If you have a website with millions of people coming through it on a daily bias, you can experiment with every single page element to find the perfect combination. If your website doesn’t have that kind of traffic you may find if you never reach any useful conclusions or you go bust waiting for the results.
Ideally you would want to have about 1000 people through a page to get a accurate conversion rate, but you would usually find by 500 people that there is a clear winner.
Based on this I suggest that you always stick to simple A/B experiments and this way you can predict how long the experiment will go on based on your how much traffic you have. So if you have 2 variations and your page is getting 100 visitors a day, you can estimate the experiment will take about 10 days, simple!

Make your pages very different!
With small amounts of traffic you are really looking for drastic increases. To do this make the landing pages as different as possible. If you are just vary something small you will see a very small difference in conversion plus it would take very little to make the whole experiment pointless, if the difference between the pages are low a couple of really keen buyer going through one page can completely throw the experiment.

Follow up experiments
If you have finished the first experiment you have two choice of follow up experiments:

1.If you find there is less than 5% difference, you need to another completely different page to compare it to. In low traffic experiments you are looking for big improvements, if the page is worth doing version testing for you really something noticeable.
2.If the conversion rate was over 15% you are on the right track. So you can either try another completely different page or you could take one element of the winning page and start altering that, so for example you could just try two completely different headers and see which one of those have the best conversion rate. The theory is the same take one element and try two drastically different versions.

Thanks for bearing with us through the rather unsuccessful first launch, we have got together some good hosting and we are now up and running on a complete new plateform. So their is lots to look forward too from the installation of our forum and the classified section. Everything is coming also any suggestion please forward them on.