Web Desktop & Mobile

UX / UI Designer

1 year and half

The Challenge

The TV5 MONDE Plus free international streaming interface was designed during the Covid-19 pandemic in a very short space of time. The team wished to audit their web interface, and optimizations were recommended via the ABtest method. 

Challenging ideas through ABtest and the alliance of parties was a lever for agile, user-centric conversion. My ability to form a trio with AB Tasty and TV5MondePlus gave us real agility in managing the various AB test projects. 

The combination of the tool’s skills, our digital and user-centric expertise, and our knowledge of the customer provided a broad and detailed vision of the scope, objectives and constraints.

Collaboration with :
  • Product owner
  • AB Tasty Support Team
  • Front-end dev
  • Project manager


A/B testing consists of putting two versions of the same web page online to determine which version is more effective.

This technique is mainly used to optimize a website’s pages for both desktop and mobile devices.

In concrete terms, we test an « A version » of the medium (which is the original version) against a « B version » (the version that involves a variation in content or form).

Use Case

Test hypothesis

Redesigning the View, Details and Favorites CTAs on the program page will make the content more attractive and visible to all users, and increase the click-through rate to a video.


UI design

The View button is available in only one design.

Original version

Variant 1

Variant 2

Test setting

The primary objective was positioned on the View button, which is directly linked to the client’s conversion KPI: « player page access rate ».

The secondary objectives were the so-called secondary buttons: Details and Add to favorites.

Click trackers have therefore been installed on all 3 buttons, including the original version, for comparative purposes.

The target is broad: the test targets all visitors regardless of their profile (new or old visitor) and country.

However, the targeted URLS are in the French language. Other languages are not included in the test.

As this is the first test of this hypothesis, we’re keeping the original version to provide a point of comparison with the version users are used to using.

So, traffic is divided into 3: 33% by variation and 1% non-tracked.

The last stage is the recipe, where all variations, including the original version, are tested to ensure that the trackers function correctly (i.e. that when a user clicks, a hit is sent and counted in the reporting).

Traffic modulation


Main Objective : 'View' button

Campagne fiable


Winning variation

Conversion rate



In view of the outperformance of the Voir button variation, we recommend using this new design to achieve the KPI objective: Increased click rate to access a video. As a first step, a patch can be put in place in target countries (French-speaking) in order to put this into production and benefit from the gains more quickly. Further development will depend on schedules and availability.

Secondary buttons : 'Details' and 'Favorites' buttons

Campagne non fiable à 100%

Detail button results
Favorite button results

Lack of reliability

This test did not reach the reliability index after more than 2 months online.

The results are therefore not tangible enough to be analyzed.

However, we can draw some interesting conclusions:

Detail button :

We’re still 1,042 conversions (clicks) short of reliability, as this button gets very few clicks, which raises the question of its usefulness for users at this stage of the journey.
It is recommended to rework the interest of the Details button as presented in the test.
A redesign project has already been launched on this page.

Favorite button :

We’re 1056 conversions (clicks) short of reliability, and this button is also very little clicked. It is recommended to rework the Favorites button and perhaps more broadly the add-to-favorites decision path in order to understand how users use it and how to improve it in order to achieve the KPI: Increase in the number of logins / account creations.