How Data Builds Brand Loyalty: What Spotify Wrapped Teaches Us About Personalisation

Every December, Spotify Wrapped shows up like a personalised digital yearbook. It tells you which artists you listened to most, how many minutes you streamed music, and what genres defined your year. Some people find it funny or embarrassing. Others see it as a badge of honour. For me, it is a curious and sometimes revealing look into how algorithms have come to shape the way we listen, and more broadly, the way we see ourselves.

As someone who works in data science and analytics, I spend a lot of time thinking about what data reflects and what it leaves out. Spotify Wrapped is a perfect case study for that. It is personal, it is visual, and it gives you just enough insight to feel known by the platform. But as I looked at my own Wrapped this year, I started to notice something deeper. Not just about the music, but about how these systems gather, infer, and present our behaviours back to us.

^ Rohan Whitehead’s Spotify Wrapped for 2024.

A Snapshot of My Listening Year

This year, my Spotify Wrapped was dominated by EDM, Indie music, Rock and whatever genre The Tech Thieves fall into. On the surface, this made sense. I like songs with catchy drops and euphoric choruses. But then I saw artists in my top five that surprised me. For example, David Kushner made the list, yet I only really listened to one song of his: “Daylight.” It went viral and I played it a lot, but I wouldn’t describe myself as a fan of his wider discography.

That pattern repeated. Several artists landed in my top five just because I had one of their songs on repeat. I didn’t follow their albums or explore their other work. And that, in itself, says something important about how recommendation systems work. They respond to frequency, not depth. They reflect patterns of use, not necessarily patterns of thought. My Wrapped painted a true picture of my listening, but not always a complete one.

I listen to music in a lot of different settings. When I am working, I often have background playlists going. When I am commuting or travelling, I pick different moods. And when I am just chilling, I go for something relaxed and immersive. Wrapped puts all of that into one bundle, compressing a year’s worth of context into one highlight reel. It is accurate in a way, but it simplifies the “why” behind the choices. And that’s where the human interpretation has to step in.

Algorithmic Memory and Personal Identity

One thing I enjoy about Spotify Wrapped is that it makes the invisible visible. It takes what we do passively, clicking, streaming, skipping and turns it into a story. Sometimes I forget how often I’ve replayed a song. But then Wrapped shows me the number of listens and I have that moment of surprise. "I knew I liked that track, but I didn’t realise it was that much."

Wrapped doesn’t just reflect our taste. It reflects how we live our lives. Songs are tied to seasons, memories, or even moods. I can pinpoint times in the year when certain tracks became personal soundtracks. And that brings a kind of emotional value to the data. It becomes a mirror, not just of behaviour, but of moments. This emotional connection is one of the reasons why Spotify’s design is so effective. It doesn’t just show numbers, it tells a story we already half-know.

But from a data science perspective, this also raises questions. Wrapped is, at its core, a filtered summary of behavioural data. It chooses what to include and what to leave out. It represents one version of the truth, based on engagement metrics Spotify considers most relevant. And like any data product, it reflects the biases built into its models.

Data Bias, System Design and User Influence

Spotify’s recommendations and by extension its Wrapped feature, are built using collaborative filtering and content-based filtering, along with hybrid models that bring in context. In other words, if you like a track that others with similar behaviour also like, you are more likely to see it again. If you frequently play songs with a particular tempo or mood, the algorithm will notice and reinforce that preference.

But here’s the thing: your current behaviour is not always the best predictor of what you want next. I don’t listen to full albums and I don’t follow specific artists with loyalty. I like songs. Often, my listening habits are shaped by the algorithm itself. That means there is a feedback loop. I click what is recommended and then I get more of the same. Spotify learns from that click, not from my reflection.

That loop also explains how an artist can appear in my top five from just one viral song. The system sees the repeat behaviour and ranks it highly, even if my broader musical identity is more diverse. In data science, this is an example of weighting bias. A small number of actions can disproportionately affect the final output. That’s fine for fun visuals, but if we think about recommendation systems more seriously, this becomes a design consideration. Are we building systems that reflect users, or ones that shape them?

And then there is the question of transparency. Most users do not know how these models work. They don’t know what counts as a meaningful interaction, or how long a song has to play before it influences a recommendation. Wrapped shows the result, not the formula. This isn’t necessarily sinister, but it does mean the user experience is curated without full visibility. If we are going to live with algorithms that tell us who we are, even in fun ways, we should have some idea of how they decide.

Reflection, Not Reduction

Despite these limitations, I don’t think Spotify Wrapped oversimplifies me. It reflects my listening habits in a fun, engaging way. It helps me remember what my year sounded like. And while it doesn’t capture everything, it gives me something tangible to reflect on. In data terms, it’s a top-level dashboard, not a deep dive.

Data-driven summaries like Wrapped can help us reflect, but they are not the full story. They are shaped by what is measurable, and that means they can never perfectly represent the nuance of personal experience. Still, they serve a purpose. They offer insight. And in a world where digital behaviour can be hard to track even for ourselves, that can be valuable.

What Can We Learn as Data Practitioners?

One of the most overlooked outcomes of tools like Spotify Wrapped is the loyalty they inspire. By turning personal data into a story, Spotify strengthens its relationship with the user. This is a form of emotional retention, not just keeping someone subscribed, but making them feel understood. When users feel like a product knows them, they are more likely to return. This isn’t just anecdotal. Studies in behavioural design show that tailored feedback loops and personalised insights can significantly increase user satisfaction and long-term engagement.

From a brand perspective, this is powerful. Wrapped turns data into a user experience moment, one that people share widely online. It becomes marketing, sentiment building and retention all at once. And it costs Spotify relatively little to implement, especially compared to traditional loyalty campaigns. The lesson for other platforms is clear: if you can make users feel seen, without making them feel surveilled, you earn more than just attention, you earn trust. If you work in data science or analytics, there’s a lot to take from tools like Spotify Wrapped. First, think about what metrics you use to summarise behaviour. Are they weighted fairly? Do they reflect long-term interest, or just high-frequency moments? Second, think about how users will interpret the summary. What story are you telling? Is it empowering or misleading?

These are not questions just for Spotify. They apply to every platform that reflects data back to the user, whether it's a fitness tracker, a learning dashboard, or an e-commerce recommendation feed. How we design these reflections shapes how users think about themselves. As practitioners, we have a responsibility to design with clarity, fairness and usefulness in mind.

And finally, we should think about joy. Spotify Wrapped works because it’s fun. It turns dry metrics into something personal. If we can build more tools that help people reflect without feeling judged and that give insight without overcomplication, then we’re not just building better systems, we’re helping people understand themselves a little more clearly.

Conclusion

Spotify Wrapped is not perfect, but it’s powerful. It shows how data can become narrative. How numbers can reflect identity. And how design choices, even in something as light-hearted as a music recap, carry deeper meaning.

As someone who thinks about data all the time, I find Wrapped to be a useful reminder. It shows the power of tracking, the influence of algorithms and the need for thoughtful interpretation. But it also reminds me that behind every dataset is a person, a mood, a moment in time.

In every webinar I give, I like to highlight the importance of data in making people feel seen. The use of our data is somewhat inescapable now, unless you throw away all your technology and flee for the mountains. Turning your client data into a fun, personal visualisation, not intrusive marketing spam, can really build brand loyalty in the long run. We know how algorithms use our data to affect our decisions as consumers, yet having fun visualisations that give back to the users in a personalised way can help rebuild trust and turn passive interaction into meaningful engagement.

Join the IoA and prepare for the future of business

Sign up now to access the benefits of IoA Membership including 1400+ hours of training and resources to help you develop your data skills and knowledge. There are two ways to join: