A/B Testing is a time-tested marketing ideology where a marketer tests the effectiveness of small changes to your marketing strategy against a “control” version of your old strategy. For online marketers, A/B testing typically works by running two versions of a website, one using your old design, and one using a slightly updated design, and randomly assigning which website each visitor sees. Finally, the website measures the conversion rate for each version of the website.
Downsides of Traditional A/B Testing
The problem with A/B testing is that it is most effective when testing small changes at a time. This can dramatically slow down the process of updating your website, especially if you want to make sweeping changes. If you change the A version too much from the B version, it will be challenging to figure out which exact changes were most effective, defeating the purpose.
As a result, traditional A/B testing typically can take months to fully optimize a website.
A Better Way: Google Content Experiments
Google Content Experiments helps avoid this by letting you simultaneously A/B test up to five different versions of a page. It’s more like A/B/C/D/E testing. This really speeds up the process.
However, it also creates a lot more complexity. Before you start testing with Google Content Experiments, create a testing plan and diagram.
Creating your Testing Plan
- Decide what changes on your site you want to test. Even with the option to test five simultaneous variants, you still want to keep your changes manageable. If possible, try to keep to changes that can be tested with a single set of metrics. For example, if you want to test one change for an increase in the purchase of a certain high-margin product, you should only include changes that affect the sales of that product in this testing cycle.
- Strategize which changes need to be implemented together. Ideally, you should only have one major change for each variant. However, some changes may need to be tested together. For example, if you want a new dynamic headline that describes images, you’ll need to also include the images. Group changes when necessary, but remember to keep it simple whenever possible.
- Diagram your testing process. Put a box at the top for each variant. Draw two arrows from each to describe what you will do if they improve your sales and what you will do if they do not. Each arrow should lead the further tests, where you combine changes or introduce additional changes.
- Set an endpoint. This doesn’t have to be the permanent end to your testing. It just has to be a point at which you have done enough testing to make some changes permanent. After this endpoint, you can introduce another testing campaign using your newer and better page as the control