It’s true that it was time to get serious about this. Better late than never. Motivation is now at its peak since a recent test has shown us everything that is possible thanks to A/B tests.
When I talk about A/B tests, I’m not just limiting myself to the functionality that Amazon offers of testing different images. We also do tests with different prices and their impact on sales.
A/B tests with prices
I’ve already mentioned this on another occasion. We started the topic of A/B tests at first by raising and lowering prices, analyzing the impact on sales, turnover and margin. In a very summarized way, the conclusion has been this: you have to raise and lower prices. It may sound too basic but that’s it. Prices have to vary. A good email data approximation is to raise or lower them every two months. We do it alternately. The price drop has an impact on the ranking, the conversion rate and consequently on the number of sales. This results in better visibility that helps when the price is raised again. The conversion rate is a little worse and the ranking is lost a little. This is compensated shortly after by lowering the price and achieving the improvement again. It is a continuous up and down that brings balance. It is neither good to keep prices high continuously nor to keep them low forever.
A/B tests with images
This test has finally opened my eyes. With something that I already knew or let’s say I intuited but oh well. In the end it is always better to have the to help developers and operators obvious in front of me so that I cannot look away. In one of the first A/B tests with images we achieved a 50% improvement. With an image that we had previously discarded because it did not seem attractive enough. Well, it beat the one we considered better by a long shot. Once again it has turkey data been shown that personal preferences are useless if they are not supported by data.
Seeing the impact
This small test we have started a recurring meeting on this topic. For us it is one of the few ways that works for us to work on important issues. We have to sit down together (well, I do it virtually from Germany) and set the tests we want to run.