It’s that time again: Spotify Wrapped is taking over our social media feeds. How many minutes of music have our peers consumed? Who’s their top artist? What’s their most played song? (We won’t say which member of HyperFinity listened to ‘We Don’t Talk About Bruno’ 73 times…).
Data science and AI in action.
Some would say Spotify is a tech company that also happens to sell music subscriptions. Spotify uses customer data REALLY well – not only to benefit their business, but also to the benefit of listeners.
Spotify Wrapped has been a viral marketing campaign for Spotify since 2016. It’s a visual snapshot of user data from the previous year: from favourite genres to the total minutes streamed.
The secret ingredient to the campaign’s success? Personalisation. The addition of personalisation means users share their results on social media – simultaneously boosting Spotify’s brand.
2020’s Spotify Unwrapped campaign prompted over 60 million shares from 90 million users (and that’s only based on data that’s trackable).
Driving personalisation with data.
Spotify takes personalisation SERIOUSLY. It’s ingrained in the customer experience they offer – from the genres and categories presented when you open the app, to your curated Discover Weekly playlist.
So, how’s it done? Spotify tracks specific data points, such as:
- Number of plays
- Added to playlist
- Customer details
Exploring the foothills of taste.
Let’s deep dive into Discover Weekly. Each week, Spotify users can listen to a playlist of new music which aligns with the genres they like, but also pushes the boundaries of their taste.
Spotify use machine learning techniques such as collaborative filtering to surface song recommendations. This technique analyses the relationships between playlists and the songs in those playlists, discovering affinity and overlap, and identifying communities consuming similar content.
But it doesn’t end there. Collaborative filtering surfaces recommendations but it doesn’t tell you WHY. Next, Spotify assigns labels (or attributes) to songs and playlists, using techniques and tools such as clustering, nearest neighbour, Natural Language Processing (NLP), audio models, and human curation. This produces attributes such as ‘Rap Caviar’, ‘Tropical House’.
The result? You can’t stop listening – and you certainly can’t unsubscribe.
But how does music apply to retail?
The data science and AI techniques used by Spotify can be applied in a retail context.
Retailers can track various data points for personalisation:
- Total spend
- Number of orders
- Browse history
- Products added to basket
- Products purchased
- Products returned/refunded
- Substitutes accepted
- Products purchased after promotion
These data points can be used to drive collaborative filtering – to surface personalised product recommendations – then attributes can be created to understand the relationship between products and consumers.
This understanding can be gathered in a ‘feature store’, for example:
Product feature store.
Customer feature store.
This foundational insight into customer behaviour can drive various personalisation campaigns. For example, emails with personalised incentives for loyal customers; mailshots curating relevant products a customer might be interested in buying; and targeted onsite media.
Driving intelligent commercial decisions.
Yep, you guessed it. We can’t write a blog without mentioning decision intelligence. But for good reason – the foundational insight created isn’t just applicable to personalisation. It also supports commercial decisions across new product development (NPD), assortment optimisation, pricing, merchandising and demand forecasting. This is decision intelligence in action.
Foundational insight also helps deliver more rewarding customer experiences – the data points collected paint a detailed picture of customer behaviour and needs. Rather than solely focusing on selling products, as a retailer can you add value to customers’ life by providing relevant advice, inspiration and partner deals?
When it’s done well, personalisation helps retailers thrive – after all, 80% of consumers want personalisation from retailers. Have you tapped into your data to provide better customer experiences? Contact us to find out how we can make your data work harder for you.