AB testing in digital marketing: how to optimize your performance continuously

Focus on AB testing, a method and technique to optimize the performance and return on investment of your digital marketing.
Are you looking for growth as far as the eye can see for your startup? Start already by monitoring your performance and seeing the points to be optimized.
For that, you will have to do A/B testing, work A/B test, eat A/B test, and even live A/B test!
But before going into pure optimization techniques, let's come back to the meaning of this Anglicism.

AB testing in digital marketing: the theoretical definition

A/B testing, also called "AB test" in French, is a method that compares the effectiveness of 2 media or digital content by testing a variant on the content or form.
This technique is mainly used to optimize the pages of a website.
Concretely, we test a "version A" of the support (which is the original version) against a "version B" (the version which implies a variant on the content or the form).
From a more technical point of view, the test phase is carried out with samples of similar audiences.
The test sample generally represents 5 to 20% of the recipient database, i.e. on a basis of 100 people, for example, 10 people test version A and 10 people test version B.
I obviously remind "that none of the guinea pigs is aware of the test" and that they, therefore, act as "normal" visitors: ie a visitor accesses a single and unique version of the support tested.
If page B obtains better results than page A (thus the original one), it is therefore declared as "winner" because it performs better.
This winning version is sent to the rest of the sample (between 80 and 95%) of recipients.
The AB Testing method is essentially based on the "Test & Learn" approach commonly used in the technology sector and more particularly among startups.

What digital media can we test?

So far, we have mainly illustrated the definition of AB Testing with the example of a test on a website page.
But we can use this method for almost all possible digital media.
Here are some examples of media that are commonly tested in digital marketing:

An email or newsletter
A landing page or a classic page (products/services)
A sponsored advertisement on Google Ads
A sponsored or "natural" post on social networks
However, this type of test can be extended to any digital medium provided it is equipped with a suitable statistical analysis tool.

What elements can be tested?

After the supports, let's now move on to the concrete elements that can be tested on support.
Let's take the example of a landing page where the idea is to test whether the variations made on it have an effect on its conversion performance, ie its ability to generate leads.
When designing the original landing page (version A) and its variant (version B), it is possible to vary the following elements:
its structure (columns, rows, position of blocks)
a button (color, text)
an image (position, size)
a text block (wording, position, uppercase, lowercase, color, link, etc.)