Spotify's 30 Song Lesson: Wrapped Rewards a Year You Already Spent, It Never Buys a New One

The first week of December, every year

Some time in the first days of December, the same thing happens on every feed you follow. A few hundred million people, most of whom have never voluntarily shared a piece of marketing in their lives, post a branded graphic about themselves. Green background, the logo sitting in the corner, a number they are either proud of or a little embarrassed by. Nobody is paid to do it and nobody has to be asked twice.

Spotify Wrapped is the most copied marketing idea of the last decade. Apple, YouTube, Amazon, Deezer, Reddit, Strava, Duolingo and several thousand smaller companies now run a year in review of their own. Every one of them is running the same play, and the play was Spotify's.

The numbers attached to it are enormous. In 2020 Spotify put it at more than 90 million people and about 60 million shares. By 2021 it was more than 120 million people. In 2025 the campaign reached 200 million engaged users in the first 24 hours with more than 500 million shares in the same day, and Spotify called it the biggest Wrapped it had ever run.

The slide that follows those numbers always says the same thing. Give people their own data and they will do your advertising for you. Your data is your distribution. It is a good line, it appears in every deck that mentions first party data, and it describes something slightly different from what actually happened.

What was already running before the first card was shared

Wrapped did not arrive fully formed in one viral December. It started in 2015 as a microsite called Year in Music, which showed you your top songs alongside the biggest tracks of the year. It was renamed Wrapped in 2016. The format that actually made it travel, the vertical story cards cut to the size of an Instagram story, did not arrive until 2017. That is three years of iteration before the mechanism everybody copies existed at all.

The audience was already there too. By the end of 2020, the year of the famous download bump, Spotify had 345 million monthly active users and 155 million paying subscribers. It added 25 million monthly users in that quarter alone, and a record 74 million across the full year.

So the campaign sold as a growth engine was running on top of a company that was already adding tens of millions of users a year without it. Wrapped is the December event of a business that grew in every other month too.

And the product underneath it is the real asset. Spotify knows what you listened to because you listened on Spotify, all year, on the commute and at the desk and in the kitchen. The data is a byproduct of a habit the company spent a decade building. Every company that copied Wrapped copied the December output. Almost none of them had the habit.

That is the date order problem for the second episode running, and it is worth saying plainly before moving past it, because the rest of this article is about something else. The thing everybody points at came last.

The decision: give the data back

Strip the graphics away and Spotify made one decision that was genuinely unusual at the time. It decided the data it collected about you was worth more handed back to you than kept.

That is not the default. The default use of listening data is to improve the recommendation engine, sell the advertiser a sharper segment, and never mention it out loud, because telling people how much you know about them is a public relations risk with no upside. Spotify took the opposite view and made the surveillance the gift.

It works because of a specific quality in this particular data. Your listening history is not a fact about Spotify, it is a fact about you, and it is flattering or funny or slightly humiliating in a way a shipping record or a bank statement never is. Music taste reads as identity. Handing someone a well designed summary of their identity, with the brand small in the corner, is not asking them to advertise. It is giving them something to say about themselves.

That is the actual mechanism, and it is worth naming precisely, because the version in the decks is wrong in a way that matters downstream. People do not share Wrapped because it is their data. They share it because it is about them and it makes them look interesting. The data is the raw material. The flattery is the product.

What Wrapped actually is

Now the part that almost never survives into the case studies, and it is printed in Spotify's own help pages rather than hidden anywhere.

You do not automatically get a Wrapped. To be eligible you have to have streamed at least 30 songs for more than 30 seconds each, and listened to at least 5 different artists, across the year. Private sessions do not count toward it. Tracks you excluded from your taste profile do not count either.

Read that as a marketing fact rather than a product one. A person who sees a Wrapped card on Instagram in the first week of December, downloads Spotify because of it, and opens the app gets nothing. There is no Wrapped waiting for them. There cannot be, because Wrapped is a summary of a year they did not spend. The campaign celebrated as the great user acquisition play of modern marketing structurally cannot deliver its own product to a new user.

Which puts the famous download bump in a different light. A 21% rise in app downloads in the first week of December 2020, measured by Apptopia, a third party estimator rather than Spotify itself, is real enough as a number. What it is not is 21% more people discovering Spotify. Overwhelmingly it is people who already had an account going to collect something they had already earned: reinstalling on a new phone, coming back after a lapse, signing in on a second device. Wrapped is a reactivation mechanic. It reads as acquisition because the graph it moves is labelled downloads.

The scoreboard

Spotify publishes Wrapped's engagement, user and share figures itself, in press materials rather than in audited financial statements, so every row above describing the campaign is company reported and none of it is independently verified. The download bump is an estimate from Apptopia, a third party app analytics firm, not a Spotify disclosure, and Spotify has never published a subscriber number attributable to Wrapped. The MAU and subscriber figures are from Spotify's own quarterly reporting and are audited. `newsroom.spotify.com` is blocked by the egress proxy, so Spotify's posts were confirmed through search and through outlets quoting them rather than by opening the primary pages.

The autopsy: what engaged counts, and the mood it cannot see

Five things are softer than the retelling.

First, the famous engagement number counts a glance. Spotify's own definition of an engaged user, in the company's framing of the 2025 record, is someone who viewed at least one story inside the Wrapped experience. One card. Not finished it, not liked it, not shared it. Opened the thing and saw a single slide. On that definition 200 million is a large and entirely real number, and what it measures is curiosity rather than delight.

This is the same substitution the last episode turned on. YouTube counted concurrent livestreams and everyone downstream converted them into people. Spotify counts one story viewed and everyone downstream converts it into 200 million people who loved it. The noun is doing the work in both cases, and in both cases the company holding the data is the one that stopped saying the careful version.

Second, the window moved and nobody marked it. In 2020 the figure was 90 million people across the campaign. In 2021 it was 120 million across the campaign. In 2025 it was 200 million inside 24 hours, and the same 200 million had taken 62 hours the year before. Those are different measurement windows wearing the same headline, which makes the series of them useless as a trend even though it is always presented as one.

Third, the shares did not scale with the users. Spotify put shares at more than 60 million in 2020 against 90 million people, and at almost 60 million in 2021 against more than 120 million. Users up by a third, shares flat. The share rate fell from roughly two in three to roughly one in two in the year the campaign was supposedly compounding. Both figures are company reported and neither is audited, so treat the exact rates loosely, but the direction sits inside Spotify's own numbers and no retelling mentions it.

Fourth, the download bump is a correlation drawn in the most crowded month of the year. December is when phones are unwrapped, when new devices need their apps installed, when holiday listening spikes, and when nearly every consumer app on earth sees downloads rise. Assigning the whole of a 21% week to Wrapped needs a control group and there is not one. Spotify has never, in an earnings call or a shareholder letter, credited Wrapped with a subscriber number.

Fifth, and this is the one that should end the argument, in 2024 the metric went up while the campaign failed. That year's Wrapped leaned on generative AI for its labels, dropped features people liked, and was received badly enough that the backlash became the story. Spotify's own executives later described it as the biggest Wrapped the company had ever had, with more negative feedback than it had ever seen. Both halves of that sentence are true at the same time, and the engagement metric only reports the first half.

That is the case against the number, and it is not a small one. A count of people who opened something cannot tell you whether they liked it, because opening it is what a person does before they find out. Spotify ran its worst reviewed Wrapped and its biggest Wrapped in the same December, and the figure it publishes could not tell the two apart.

What first-party data actually is

Now the test this series almost never gets to run, and this time it runs itself, because everybody copied this one.

If the lesson were that your data is your distribution, the same data in another company's hands should travel the same way. The comparison is sitting there. Apple Music has run Replay since 2019. YouTube Music has Recap. Amazon Music, Deezer and Tidal all publish a year end summary. Reddit has Recap, Strava has Year in Sport, and Duolingo has a Year in Review the company is genuinely good at.

Apple Music Replay is the cleanest control group this series has been handed. Same category, same underlying data, a larger and richer company, a subscriber base in the hundreds of millions, and a summary of your listening year delivered every December. It does not travel. Most Apple Music subscribers do not know it exists, and the ones who do largely treat it as a playlist rather than something to post.

So the data is not the asset. If it were, Replay would work. What separates them is everything around the data: a format built for a vertical story instead of a webpage, a fixed date that lets an expectation build all year, a design with enough personality to be funny, a decade of iteration on which cards land, and a brand young enough that posting it says something the poster wants said about them.

The honest version of the lesson is narrower and more useful than the slide. First party data is not distribution. It is raw material for a gift, and the gift only travels if the thing you hand back flatters the person holding it and is shaped for the place they are going to put it.

There is a second condition, and the eligibility rule is what exposes it. You can only hand somebody their year if they gave you a year first. Wrapped is a receipt for a habit. A company without the habit has nothing to summarise, which is why so many copies are padded with numbers nobody asked for: how many invoices you sent, how many times you opened an app, how many minutes you spent inside a dashboard. The data existed. The year did not.

And the vanity metric for this entire category is opens. The clarity metrics are the ones nobody screenshots: what share of the people who opened it reached the last card, what share shared it, and whether the people who came back in December were still there in February.

The right order: earn the year, then hand it back

Start with whether you have a year worth giving back, because most companies do not. The test is blunt. Is there a number inside your product that a customer would be pleased or amused to see about themselves, and could not have got anywhere else. Not a number you are pleased with. A number about them. If the honest answer is no, a year in review is not your campaign, and running one anyway produces the padded version that makes a company look like it has been measuring its customers rather than serving them.

If you do have one, design for the surface rather than the report. What made Wrapped work in 2017 was not more data than 2016, it was cards shaped like the place people were already going to put them. Build the artefact for one specific surface, at the size that surface uses, and cut every number that will not fit on it. A PDF summary is not a Wrapped. Neither is an email.

Name the clarity metric before you build it, because this category is unusually good at producing impressive numbers that mean nothing. One metric, one window, one owner, written down while the idea is still a draft. Opens is not it. Completion rate, share rate, and the 60 day retention of the people who came back are three that would actually tell you whether it worked, and all three are harder to put in a press release, which is precisely why they are the ones to pick.

In Saudi Arabia this play is closer to reach than most of the ones in this series, and the reason is structural. The customer base here is young, concentrated on two or three platforms, and unusually willing to post about itself, which is the exact condition Wrapped needs and the hardest one to manufacture. The constraint is not the audience, it is the material. Most mid market companies here hold a year of transactions with no personality in them, and a summary of somebody's deliveries or invoices is not identity. So settle one question before anyone designs a card: does the product generate a fact about the customer that the customer would repeat about themselves. A gym, a coffee subscription, a bookshop, a driving app and a fantasy league all do. Most B2B does not, and that is a useful thing to know in the week before December rather than the week after.

The takeaways

Wrapped cannot acquire the user it is famous for acquiring. Eligibility needs 30 songs at more than 30 seconds each and 5 different artists across the year, so a person who downloads Spotify in December because of a Wrapped card has no Wrapped waiting for them.

An engaged user is someone who viewed at least one story. The 200 million is a count of glances, and it is the same substitution the Red Bull episode turned on, with one noun quietly traded for a better one.

The window changed and the headline did not. 90 million across the campaign in 2020, 120 million across the campaign in 2021, 200 million inside 24 hours in 2025. Three different measurements presented as one line going up.

Shares stayed flat at about 60 million while users rose from 90 million to more than 120 million, so the share rate fell by roughly a third in the year the campaign was growing fastest.

Apple Music has run Replay since 2019 on the same data in the same category and nobody posts it. That is the control group, and it says the data was never the asset.

In 2024 Spotify ran its most criticised Wrapped and its biggest Wrapped in the same week, and the metric it publishes could not tell the difference.

Frequently asked questions

This article is part of BMD's marketing case-study series. Episode 12 is about what a data campaign can and cannot do for a company. Episode 13 is a 4,500 dollar video a razor company used to take on Gillette, and the subscription underneath it that did the actual work. That is Dollar Shave Club.

Sources and further reading

About BMD

Most companies don't have a marketing problem. They have a marketing department that was never built. BMD is a boutique consultancy that installs structured, measurable marketing departments inside mid-market companies across the GCC. We don't run your campaigns, and we don't hand you a strategy deck and leave. We build the operating system: the structure, the measurement, and the ownership that turn marketing into a function leadership can rely on. The method is the BUILD framework, published and practiced: a book, an online program, a community of Gulf founders and marketers applying it, and diagnostics that replace assumptions with measurement. Delivered in Arabic and English, founder-led.

Redha Alayesh

A marketer with a software engineer's discipline and a scientist's mindset. Across 40+ marketing departments in the GCC, he built the BUILD framework to solve the problem he kept finding: capable marketers trapped inside companies that never built them a department.