From A to B of A/B testing

The point of A/B or ‘split’ testing is to identify the most successful content and pricing. It allows you to optimize your listing, using real input from your customer base.

We call it A/B but ‘split’ testing can be used to try out more than just two possibilities. You could try half a dozen.

To do this, you decide on a number of alternatives that you’d like to test. For instance, your product description might be “Pink neon marker pen makes your writing glow in the dark,” but you might want to trial “Glowing pink neon marker pen…” or “Shocking pink…” By doing split testing you’ll be able to see which of these options works best.

But when you’ve got a few products and a half a dozen features each that you want to test, it gets cumbersome. That’s why there’s an A/B testing feature in CashCowPro. You enter the variables you want to test, and the software does it, and shows you the sales results.

Amazon gives you sales data by day, so it’s important to change the variables being tested at midnight. That’s another reason, unless you’re a real night bird, you probably want to use an automatic testing system rather than do the job manually!

As well as testing your product title, you can test your main image and three or four others, and you can test your product description. Testing which photo works best as lead takes the subjectivity out of the equation – you may personally not like product photos with a white background, but if that’s the one that sells, you’re going to use it.

Also, and this has a huge impact, you can test your sales price. So for instance if you’re selling at a premium price, and the entry of new competitors has seen your sales slow significantly, you might want to test what happens if you drop your price. You could test dropping your price all the way to your competitors’, keeping a ten percent premium, or only going half way to meet them. If it turns out that a relatively small drop in price gets your sales motoring, then you don’t need to drop it all the way. By split testing, you’ve saved yourself a whole load of profit you’d otherwise have lost.

You can also test which set of five features gives you the most sales. You may think you know your customer base pretty well, but sometimes after four reasons why you’re the very best, they want to know a particular hygiene factor – such as “easy to clean” (a hygiene factor in both senses!) or “child-safe”. Just adding that might make a big difference. Equally, sometimes switching the top two benefits might make a difference. Split test the variations, and you’ll be able to see whether you can increase your sales with some simple changes.

You can also decide how you want the test to run. The problem with a 24 hour switch is that if you have one variant tested on Sunday, and the other on Monday, you may be comparing apples and pears. People are at home and shopping on Sunday – they’re at work on Monday. So it’s not the best test for consumer goods. The other way to run the test is to run your variants for a week at a time till each one’s been tested. That will take longer, but might give you more reliable results.

If it’s this simple, where does your particular expertise come in? In having beautifully crafted, Amazon-compliant titles to test, or in having accurately identified at least seven features that will sell to your customer base. Split testing just means you can find out which is the best. You still have to do the hard work yourself!

And when is it finished? Well, the right answer is ‘never’ – but there is a law of diminishing returns. If you were seeing 10-20% or more increase in sales for every change, and now it’s down to a couple of percent, your time might be better spent on sourcing new products.


Leave a Reply

Your email address will not be published. Required fields are marked *