Améliorer l’expérience utilisateur chez Deezer – Interview de Tom Abourmad

Within the streaming platform, the evolution of the user experience is not solely dictated by the analysis of usage data. A vision that, on its own, masks the needs or intentions of users.

He publishesImproving user experience through the use of usage data? Of course, but not only, argues Tom Abourmad, User Researcher within Deezer’s music streaming platform. Research on usage data stops at why. However, in order to imagine the features of tomorrow, it is necessary to be able to identify the needs of users, their intention, indicates the specialist, a member of a team of four people specialized in user behavior research, a team itself integrated into the product/technologies department of Deezer.

To guide the product teams, Tom Abourmad has various methods, what he calls his toolbox. From traditional focus groups, to questionnaires in the application, through moderated or non-moderated interviews, tests or longitudinal studies. Not to mention the wealth of spontaneous feedback from users, through the reviews they write, the feedback to customer service, or experience trackers, adds the User Researcher. These are based on series of questions submitted to users at key moments (for example, when they are about to cancel their subscription), allowing for a longitudinal follow-up of the experience lived by users.

In-app studies: valuable, but to be used in moderation

To manage these campaigns, Deezer has gradually generalized the use of the SaaS platform Qualtrics, an American company specialized in the study of consumer behavior online, once a part of SAP. The deployment began about two years ago, covering the consolidation and analysis of rather advanced questionnaires, according to Tom Abourmad. Since then, the software has been extended to experience trackers, but also to in-app studies, very short surveys injected directly into the application in a specific usage context. It is an optimal source of knowledge for product managers and it reduces the solicitations we receive for ad hoc studies. The difficulty lies in the fact that these insertions must remain very light for users, indicates the UX specialist.

The deployment of Qualtrics has finally allowed the User Research team to follow in the footsteps of the data analysis team. This team produces many dashboards for users to consult the data on their own. We were able to adopt the same approach, on qualitative data, observes Tom Abourmad. An important point according to the latter. After proving the value of our analyses internally, we now face a multiplication of study requests, continues the person in charge. Like the Data Office, the User Research team therefore relies on the increasing autonomy of users in the analysis of data to absorb this growing demand.

He publishesImproving the Flow feature

The interest of the research approach particularly appeared on the so-called Flow feature of the streaming application. A central feature of Deezer that aims to answer the users’ first question: what am I listening to? The Flow should allow users to quickly play music that they will like, summarizes Tom Abourmad. Widely used, the function generates a mass of usage data. Except that this monitoring yielded as many positive feedbacks as potential improvements, recalls the expert. The intervention of the team specialized in UX research then makes it possible to highlight that the function, which aggregates favorited tracks and suggestions from the recommendation algorithm, actually meets various use cases.

Tom Abourmad, User Researcher at Deezer: AI is obviously called upon to play a role, especially in digesting the data collected by the application. (Photo: D.R.)

Our exploratory research has shown the importance of the user’s mood, the environment in which they are immersed at the time of using the Flow, continues Tom Abourmad. This finding leads to an evolution of the function, with an option allowing the user to configure their Flow according to their current needs, by guiding them towards more musical discoveries or, on the contrary, towards more replays.

The Deezer teams are currently focused on AI. This technology is obviously called upon to play a role, especially in the digestion of the data collected by the application, assures Tom Abourmad, who observes however that AI will not magically solve all questions around data interpretation. According to him, generative AI in particular will bring value internally, for example to synthesize user feedbacks. As for its integration into the application itself, if Tom Abourmad recognizes a real interest in the technology to improve search functions, a lot of Data Science work remains to be done, he acknowledges. It would be necessary to associate moods, genres, and other characteristics to around 100 million tracks in order to be able to propose relevant tracks in response to a prompt, he says. If the potential and the data are there, it remains to seamlessly integrate this type of functionality into an application that currently has around 10 million subscribers.