The Story of Spotify Personas

Here at Spotify, we regularly ask ourselves whom we’re designing for. And since listening to music is so universally standard, it might sound at first that the reply is ‘everyone’. After all, Spotify is obtainable as a free and paid product. It can be utilized by anybody with a telephone, pc, automotive, set of sensible audio system, or many different units. It’s current in over 79 markets and it presents experiences – like Daily Mixes – which are personalised to each single listener.

Yet designing for a mass, generalised viewers isn’t more likely to find yourself pleasing ‘everyone’. So in 2017, our group was challenged to create a greater understanding of present and potential listeners. We needed to agree on easy methods to differentiate the wants of those listeners and the issues our merchandise may clear up for them. We wanted an answer that was sturdy and versatile sufficient to work for autonomous groups, figuring out of various workplaces, in numerous international locations and on totally different elements of our merchandise. And we have been decided to place a face to our listeners – an id that everybody at Spotify may recognise and speak about with ease.

We responded to this problem by designing personas.

How did we craft the personas?

User-centred design has a number of faculties of thought on how finest to create and use personas. The common concept is that capturing and clustering the wants, targets, habits, and attitudes of present and potential customers helps to construct a strong understanding of the issue area. For us, our personas software is an instance of a boundary object – a sturdy and dependable artefact that’s versatile sufficient to encourage discussions, share info, and adapt to the wants of the product improvement course of. And we developed it in two phases:

Phase 1 (2017)

In Phase 1, we scoped our evaluation to US listeners. We picked this market on account of its measurement and the number of listening behaviours that emerge from the lifestyle there – for example, lengthy commutes, suburban existence, and so forth. At the beginning, we mentioned the thought of clustering behaviours gathered from our present knowledge. But we moved away from this method as a result of it revealed solely superficial information about our listeners and hid the explanations behind their behaviour. It additionally failed to assist us perceive why potential clients hearken to music. So as an alternative, we determined to check listeners of various ages, incomes, household sorts, existence, music cultures, and extra. We used a mix of diary research and contextual inquiries to gather this knowledge.

Early within the evaluation, we observed that individuals’s wants or causes for listening to music have been constant, even in numerous clusters — that’s, to kill boredom, to really feel productive, to entertain themselves, and many others. But what was totally different was their angle in direction of music consumption, the worth they noticed in paying for music and their behaviours round units in numerous contexts.

As a consequence, we dominated out the thought of clustering primarily based on wants alone and used a mix of Alan Cooper’s methodology and the Grounded Theory method to construct our personas as an alternative. We transcribed our interviews minute-by-minute. Then, we coded and clustered them into wants, attitudes, machine habits, contexts, and different dimensions with a view to determine one of the best cluster combos. Two instruments — Mural and Airtable — have been significantly helpful throughout this part.

Phase 2 (2018)

In our subsequent part, we constructed on a key Phase 1 perception – that relating to music listening, context issues. Sure, there’s worth in creating summary dimensions, reminiscent of wants and motivations. But in the end, individuals use Spotify in the true world. Their machine ecosystems, bodily and psychological talents, and different contextual components form their listening selections. And so, combining the learnings from Phase 1 with a literature evaluation of theories from sociotechnical techniques and adaptive computing, we determined to focus Phase 2 on how individuals hearken to music collectively.

In this part, we sought to unpack the nuances and complexities that come up when individuals pay attention collectively at house, within the automotive, with youngsters, and extra. And since this work constructed on our earlier analysis, we as soon as once more stored our sampling inside the US. We included roommates, empty nesters, companions with and with out youngsters, households with toddlers, youngsters, and others. Our objective was to make sure we had an intensive number of conditions the place individuals got here collectively to hearken to music.

Unlike in Phase 1, we adopted up our diary research and contextual inquiries with a bottom-up evaluation utilizing the Grounded Theory Approach. Qualitative coding revealed insights that we might have in any other case missed and resulted within the Listening Together Framework™, our software to speak the outcomes to a broader viewers.

While individuals may need the identical issues or wants, the present habits decide the present strategies they use to handle these issues. Attitudes decide how totally different individuals will undertake merchandise designed to satisfy their wants.

Next, how ought to we characterize our listeners?

Representing personas poses a difficult problem: we would like them to be relatable, however they’re not 1:1 matches with actual individuals. Believable human traits and flaws assist create empathy with issues and desires. But we do not need teams to be wrongly excluded primarily based on the traits we have picked. So discovering a steadiness is an important step if we’re to create helpful and plausible archetypes.

For that purpose, we arbitrarily picked genders, names and appearances that matched the vary of individuals we interviewed. While personas exist independently from these traits, they have been basic to make them memorable as individuals. And deciding which human traits to incorporate in every of the personas was particularly difficult. To achieve this, we decreased the illustration of personas to key phrases, colors, symbols, and power ranges reflecting their enthusiasm for music. This train helped us navigate by the variations of poses, facial options, clothes, and visible kinds we created.

To steadiness out these particular traits, we used flat illustrations with our model colors, giving them a extra summary look. Avoiding a too-realistic illustration made the fabric simple to refresh with evolving illustrative kinds. It was additionally a lot simpler to breed in excessive or low constancy, since sketching a selected pose or selecting a color palette can be sufficient to check with a persona.

How did we share our work?

We didn’t wait till our personas have been full earlier than sharing them – we really began interested by communication as quickly as we started our analysis. We spent loads of time testing our asset concepts in pilot workshops. The objective was to combine with our present practices seamlessly. And by following our group wants, we crafted a communication technique for Personas that features digital property, bodily property, and workshops.

Digital property

Traditionally at Spotify, we create Google shows when reporting again analysis – and generally, these get misplaced amongst all the various different shows produced! But this time round, we envisioned our personas work to be related for not less than a few years. So we created an interactive web site, shared throughout Spotify workplaces by bulletins and posters. Having a digital supply of fact for the analysis was particularly helpful each time we wanted to replace the examine or add new learnings.

Physical property

Raising consciousness concerning the personas was helpful, however we did not wish to cease there. We needed to create enjoyable, playful methods for the groups to include them into their workflows. So we created property that groups may use on their very own, whether or not they have been working one-hour mini-workshops or design sprints over a number of days. These property have been made obtainable by our personas web site.

Workshops

One of probably the most highly effective modalities for studying that emerged throughout our pilot workshops was ‘learning by doing’. So the person analysis group hosted workshops with product groups and helped them to make use of personas in a approach that was related to their particular areas.

What was the impression?

Since our groups are so autonomous, we realised proper from the beginning that the personas can be related to all of them in numerous methods and at totally different phases of their work. For that purpose, nobody was mandated to make use of personas. Yet, as a dependable, sturdy, and punctiliously designed info artefact, we’ve seen many groups past the product organisation undertake them into their work and vocabulary over time – together with these throughout Marketing, Content, and Brand.

For occasion, groups that wish to create options from scratch can now select their personas, map out the present alternatives, decide a course, and begin ideating from there. Although personas don’t exchange person analysis, they might help us create educated hypotheses and save us time – which means we don’t have to run foundational analysis each time we wish to discover a brand new matter inside the music listening expertise. Our groups can now focus their assets on diving deeper into issues from the extent set by the personas.

Equally, when groups are extra targeted on sustaining options, they’ll now map out their work and see how totally different personas would use it. They can create psychological mannequin diagrams for various personas and uncover how they expertise their journeys. And in doing so, they’ll refine the options to raised match sure methods of listening to music, whereas ensuring they don’t alienate others.

Crucially, the personas are slowly changing into part of our inner vocabulary – a method of serving to groups to pick and determine which methods of listening are being affected. We can’t optimise a function for each single one in all our listeners. So right this moment, it’s frequent to see groups having their product roadmaps centred round particular personas as an alternative.

A protracted course of, with long-lasting outcomes

Sometimes, with a view to transfer quick, it’s a must to transfer gradual. Foundational analysis initiatives, like the event of personas, take time and are resource-intensive. Yet the learnings profit us lengthy into the longer term. And listed below are only a few of them:

• When unsure, over-communicate. We want an everyday cadence to share particulars and progress across the organisation – this may add overhead, but it surely ensures alignment and transparency. We used Facebook Workplace, Slack, and emails to maintain the stakeholders up to date all through the method.

• Keep your disciplines shut. Our course of needed to transfer shortly from behavioural evaluation to fieldwork, then straight onto asset creation and scoping wants, attitudes and habits, by the usage of surveys. The pace we moved was solely attainable by having design, person analysis, and knowledge science built-in all through the method.

• Know your viewers. Adopting new frameworks could also be a big change for some product groups. So we spent a lot of time attending to know their workflows, working pilot workshops, and welcoming them to fieldwork periods with a view to construct belief and scale back any potential resistance to vary.

As Spotify continues to develop, we count on to increase and adapt our personas for markets exterior the US, in addition to broadening out our space of examine to additionally embody podcasts. There are thrilling instances forward and many extra work to be carried out – we’re wanting ahead to the subsequent chapter within the story of Spotify personas. 🙂