carlos-muza-84523-unsplash

How to test, measure & optimize texts large-scale

Have you ever thought about changing the way your target group is addressed in the texts on your website or in your shop? Then you have certainly thought about how to reach your target group even better and how they will feel even more addressed.

Have you ever thought about changing the way your target group is addressed in the texts on your website or in your shop? Then you have certainly thought about how to reach your target group even better and how they will feel even more addressed. That’s the right thing to do!

After all: If you want to run successful websites or shops on a permanent basis, you always have to stay up to date and continuously try out new things. But how should we know what works and what our target group is attracted to? The answer is as simple as it is logical: test, test, test – and measure!

A/B tests with web texts – a lot of effort, little benefit?

If one thinks of classical A/B tests, these mean an enormous amount of effort, especially in text creation: First of all, two versions of the same text have to be created editorially in order to test them against each other – all other components should be the same. If you then have test results which of the two variants works better, it is a great effort to optimize all existing content according to the new findings.

This is where automated text creation comes into play: It allows you to make targeted changes to the text concept and then automatically generate all the texts again.

Set up measurable tests and implement measures large-scale

The process behind it is quite simple:

  • Create two versions of a text (e.g. a product text with a loose, direct approach to the reader and a product text without a direct approach).
  • Build two landing pages that differ only by the product text, or use two comparable products in the shop for the two text variants.
  • Define a target (e.g. sales) and measure which variant converts better.
  • Adapt the text concept to the better converting approach and then generate texts with this approach for all products.

You make the changes once in one place and the effect multiplies. In the same way, you can, for example, test the use of call-to-actions in the text, change the sentence order within the text, try out the effects of formatting such as boldings, paragraphs, or subheadings, and so on.