Cyrillus is a French ready-to-wear brand for the family that was created in 1977. I have been working with the marketing teams since 2014 to improve site performance and user experience. At the beginning of 2015, the redesign project was accepted and creation began several months later with altima°.
To guarantee ROI for the redesign, we adapted our testing methodology to favor information gathering. Some subjects were divisive among the internal teams and testing made it possible to make quick decisions (product plot, filter positions, etc.).
For example, what order tunnel structure to adopt? Should the order summary be immediately visible for the customer or should more room be given to the products? These changes present risks so we made a quick test on the former site:
Placing the summary at the bottom of the page reduced per visit value by 8% across the entire tunnel and by 14% for new customers. In short, it didn’t please anyone so we quickly withdrew this idea from the redesign!
Several points could not be tested before production start and Caroline, the web project manager, came to us with this brief: “Further to the site’s redesign, we would like to check the impact of the new header on sales and user behavior. Can we question it without changing everything?
The header was one of bold changes made with a modern but non-conventional structure:
The redesign met several of the brand’s needs in a limited space:
The reassurance arguments were presented in the central area of the pre-header. The size of the area was limited and so to fit the 3 arguments, an animation was created to move from one to the other. The advantage of the animation is to draw the user’s eye but then s/he must wait approximately 4 seconds to see the 3 arguments (sic, too long?).
The new advertising area was displayed for a first visit or opened to the bottom with a click on the yellow tab. Its height was important and this novel behavior was far removed from web conventions. It could come as a surprise and we asked ourselves what its impact on new users (bounce rate?) or the offer presented (ROI) might be:
This was a lot of information in a small area and I had several questions concerning conversion:
To answer all these questions, with the Cyrillus marketing teams, we prepared A/B testing with 6 combinations.
To test the impact of the advertising area, we deleted it completely. That was simple and we rapidly knew what the impact on gift cards was.
Another variation pushed the current offer in a set banner with no animation. In this situation, we measure the impact of the animation and the visibility of the gift card:
Simultaneously, we tested the placement of the reassurance and its handling. In this situation, it was highlighted with a red background and the animation between the arguments which was potentially bothersome, was deleted:
We also tested the placement of the reassurance below the browser bar which is more conventional and has the advantage, in our case, of adding value to the browser. We made 2 variations, one black:
and a much more visible yellow one:
To finish, we tested a personalized version of the last 2 variations. We did not know if the arguments presented were relevant and whether the talk should be personalized according to user type.
New users are exposed to a message that pushes a seemingly dedicated offer while former users keep the initial reassurance text. In fact, the arguments are identical in both situations only the wording changes:
After 2 weeks of testing, the black personalized version stood out from the pack to reach +10% per visit value across all profiles. Without the personalization, the variation had no impact. That lesson was precious and very interesting.
Another interesting lesson, the color yellow did not have the expected effect and this variation was slightly negative overall. For new users, the optimization was only +1.8% where the black version reached +11.6%! These graphic effects had little interest here and probably on newsletters and other advertising aids.
The other combinations were slightly negative and stable over time. Gift card sales did not evolve and this zone is of little interest. As a result, I would recommend deleting it to find a better area for advertising.
This test answered numerous questions on artistic and UX choices which were made during redesign. Originality can be dangerous and this testing methodology calms ardors.
Personalization has a positive impact and the 10% increase in per visit value across the entire site is huge! It would be interesting to test other approaches. For example, is it possible to personalize advertising for certain important operations, the enrollment process, etc.?
Can you safeguard your redesign with A/B testing? Is your advertising personalized?