Ideas for A/B Tests from Ebay…

We can definitely learn from websites such as Amazon, Ebay, Apple and others with huge resources (money, time etc…). Of course, they are carrying out A/B testing on their product pages and across their website to get the best results and there is no reason why you can not take advantage of this.

Now, I’m not saying that you should be copying everything that Amazon are doing. However, you can certainly do some A/B testing based on some of the features or design of sites such as Amazon and Ebay.

We visited Ebay recently and I have to say I liked some of the design and could see how it would encourage a conversion. We liked it so much we tried some of the methods on a client site.

Let Users Know the Product Qty

Typically on e-commerce sites, there is no problem with letting users know the product quantity. If it is high then users will know they have more time to do research whereas if it is low then users may well decide to buy asap as they can’t afford to miss out. I like the way on the Ebay basket page it overlays the image with a banner.

Results: We implemented this ‘Last One’ banner feature on a client site and our A/B testing results showed a fantastic increase within the first 2 months of 18.2%! It should be noted this was a client with a lot of seasonal products, therefore nearly 30% of the products on the site would be the last one.

Important Info Above the Fold

Usually it is common for online retailers to have the important information on the product page. In terms of important info, I’m talking delivery, returns and payment. However I commonly see sites where all this information is below the fold or hidden behind tabs. I do like Ebay’s approach of providing a short snippet of this info above the fold with a anchor link to further details.

Results: We condensed snippets of the important info into a table which was above the fold (for over 82% of the site’s users (source: Google Analytics)) with anchor links to more detailed info. We seen an increase in the conversion of 5.2% from the tests over a period of 3 months with in excess of 150,000 users taking part in the tests.