Tag Archives: Spotify

How Spotify’s “Wrapped” Spoke to Me

December means lots of things. End of the year, colder weather, the holiday season, and another edition of the “Wrapped” playlist that Spotify curates for its premium subscribers.

Spotify aggregates the listening data it has on you to present your very own year-end listening recap with a delightful, animated accompaniment. In previous years, even though data is data, for whatever reason I’ve felt Spotify missed the mark on how I used the platform.

Spotify Wrapped

But this year, they nailed it.

Spotify confirmed my wildly diverse tastes by indicating I’ve listened for 24,350 minutes to 1,081 artists in 51 genres in 2023, with Rock, Pop, and Hip Hop at the top. My recent obsession with Noah Kahan and Brandi Carlisle apparently puts my listening tastes in sync with those residing in Burlington, Vermont. “Flowers” by Miley Cyrus was my most played song. What can I say? I can love me better than you can.

My most played artist was Crowded House with a peak listening month of May, which makes total sense since that’s when I saw them play live in my town.

My favorite part of the Wrapped video is a new feature, introduced with “Hold up, someone’s on the other line”. It was Crowded House lead singer Neil Finn thanking me for being a fan and announcing a new album is coming out next year.


Neil Finn Crowded House

Crowded House’s Neil Finn sent me a video message on my personalized Spotify “Wrapped”. Photo credit: Ben Houdijk/Shutterstock

The Wrapped section of my Spotify dashboard has an already curated and personalized Top Songs of 2023 playlist, other artist video messages, which of those artists are coming next year, and some available merch to buy for the holidays. It isn’t 100% perfect…my colleague David tells me his Wrapped alerted him to an upcoming concert by one of his top five artists…who died in early September. Oops.

“Wrapped” is intensely personal, intentionally viral, and a powerful motivator of brand loyalty.

It’s easy to dismiss how applying the lessons of Wrapped could work for your brand. Of course, you don’t have Spotify money. You don’t have the resources to create gorgeous content at a Wrapped level of graphics.

But, you may have access to a robust database of consumers. Hopefully, you’ve collected some information on these consumers. At a basic level, you may have some demographic information, but hopefully you’ve conducted some surveys to learn more about them. If you work at a radio station, do you have information about who your listeners’ favorite artists are or what shows they like or what contests they’ve played? Do you know what they do for a living and when they listen most often?

Knowing you probably don’t have the access or algorithms that Big Data provides like Spotify does, 1) what information can you glean from your database and 2) what can you do with it?

One thing most radio stations do have is talent access, including the artists played on the station and the talent on-air. What personalized content can you generate for your audience to help build brand loyalty and encourage sharing via social channels? Local radio stations also have the power of community, focusing on elements germane to that specific audience.

What are some other ways you can make the listening experience more experiential and less transactional?

As you consider planning for 2024, think about ways your radio station or other media brand can deepen bonds with your audience, often in ways that are low or no cost, with data you already have or are easy to obtain just by asking.

Retention Rate: The Magic New Metric for Measuring Hit Potential in Streaming Data

Today’s Tuesdays With Coleman blog post comes courtesy of Matt Bailey, president of our sister company, Integr8 Research. Integr8 Research is focused solely on new music research, and Matt regularly offers thought leadership to his clients regarding the interpretation of streaming data. In this entry, he explains Retention Rate, a new metric to recognize which songs have staying power.

Within a week of its debut, Spotify users in the U.S. played Taylor Swift’s “Anti-Hero” a record-shattering 31 million times—more streams in one week than any other song at the time.

Naturally, CHR and Hot AC stations also immediately added the highly anticipated song by an iconic core artist.

Surely, with record-breaking streaming figures and substantial airplay, “Anti-Hero” was destined to be the most popular song of 2022, as validated by stations’ callout research, right?



In this post…

  • Why record-breaking streaming doesn’t mean big hits for radio
  • Introducing Retention Rate
  • Streaming peaks reflect the artist’s power; Retention Rate reflects the song’s appeal
  • Practical ways to use Retention Rate



While “Anti-Hero” is hardly a flop, it also hasn’t been the instant home run that “Bad Blood” or “Shake It Off” were. On average, about 35% of CHR and Hot AC listeners said they loved “Anti-Hero” at its peak.

That’s strong enough for a secondary current, but the biggest hits typically see 50% of listeners exhibit strong passion for them using Integr8 New Music Research’s methodology. Harry Styles’ “As It Was” and “About Damn Time” from Lizzo both achieved this metric among our typical clients.

When we examine how these bona fide radio hits performed during their peak week in the U.S. on Spotify, there’s seemingly no correlation at all between streaming plays and passion among radio listeners: “As It Was” had a strong peak week, but Lizzo’s “About Damn Time” maxed out at well under one-third as many streams during its peak week, as did “Anti-Hero.” We will demonstrate this in a bit.

As we’ve noted in several previous posts, streaming data amplifies active fandom, which means a song’s peak week on streaming will often be on or shortly after its debut week. As weeks go by, however, streaming data increasingly reflects the passive fandom of people who discovered the song after its release.

With those behavior patterns in mind, let’s fast-forward to “Anti-Hero’s” 10th week from its peak. (Why 10 weeks? Typical top 10 hits for CHR radio reach their peak passion level among listeners around 10 weeks after debuting.)

Taylor Swift - Anti-Hero

Spotify users in the U.S. played “Anti-Hero” about 5.7 million times ten weeks after its record-breaking debut week. That’s still sufficient to be among Spotify’s Top 10 songs in subsequent weeks. However, that’s less than one-fifth as many plays as its debut.

In contrast, Lizzo’s “About Damn Time” only garnered around 8.5 million streams when it peaked in its fourth week. Fast forward 10 weeks, and Lizzo’s summer anthem was still pulling almost 6 million streams.

Lizzo - About Damn Time

That puts “About Damn Time” on equal footing with “Anti-Hero” in its 10th week after its peak.

However, simply examining how many times people play a song at 10 weeks doesn’t explain why “About Damn Time” was a much bigger radio hit than “Anti-Hero” has been, either. There’s still no clear correlation between streaming plays and callout passion scores.

If you can’t simply examine how a song performs at its peak and you can’t examine how a song performs once the initial interest in the song dies down, is there any streaming metric that does track with listeners’ passion for a song, as measured in your callout research?

Comparing plays for a song’s peak week on Spotify with its play count 10 weeks after its peak reveals a telling story.


Retention Rate is the ratio of how many times streaming music consumers streamed a song in its 10th week after peaking, compared to how many streamed the song during its peak week.


Comparing the Spotify streaming retention rate for the songs for which we initially examined passion in callout research, we finally begin to see some semblance of a correlating pattern:

Why does Retention Rate make sense?


A song’s peak streaming week is a measure of how many fans an artist has and how passionate and engaged those fans are. Artists like Harry Styles, Morgan Wallen, Drake, and of course Taylor, have fan bases that run wide and deep. The decision to stream a brand-new song when it drops has everything to do with the artist and nothing to do with the song itself.

In order for a song to continue to garner strong streaming plays, at least one of two things must happen—and generally both:

1) Those diehard artist fans who played the song when it debuted must decide they like it enough to keep playing it week after week.

2) Music fans beyond the artist’s diehard supporters must discover and like the song enough to start streaming it.

For music consumers who like a song but aren’t superfans of the artist, they have to discover the song out in the wild because they’re not explicitly seeking it out: A friend has to tell them to check it out. They have to hear it on a curated streaming playlist. Or perhaps—just maybe—they hear it on the radio.

Usually both #1 and #2 must happen for a song to become a lasting hit.

If an artist’s biggest fans are underwhelmed, it’s unlikely listeners who aren’t even that into the artist will be big fans of the song.

Even for superstar artists like Taylor, there are generally a lot more people who love the song than there are who are hard core fans of the artist. They may “like” the artist, but they’re not buying concert tickets, t-shirts, and fan club memberships.


Retention rate captures both phenomena. A high Retention Rate means:

  • An artist’s biggest fans like the song enough to keep playing it week after week. A low retention rate means those fans may still love their artist, but found the song itself a letdown.
  • Retention Rate also reflects new people who have discovered a song after it debuted and have gone to Spotify to play it. A low Retention Rate can indicate few people are discovering the song passively.

In short, while a song’s debut streaming plays is a referendum on the size and passion of an artist’s fan base, a song’s ability to remain among the most streamed songs week after week is more predictive of a song’s hit potential.


First, forget about using Retention Rate exactly as we have described it here. Nobody has time to track weekly streaming data for dozens of songs week after week in Excel to precisely calculate a 10-week retention rate.

Heck, I don’t have time to do it, but I find it too much fun to resist.

Fortunately, you don’t have to spend hours creating Excel formulas. Here are some easier signs to examine when you’re reviewing streaming data:

1) How do the song’s streams hold up in the first couple of weeks after debuting?  Harry Styles’ fans in the U.S. played “As It Was” on Spotify over 24 million times the week it debuted. As expected, the song got fewer plays in subsequent weeks—but with 17 million streams in its second week, “As It Was” maintained 69% of its weekly plays in week two. In its fifth week, “As It Was” still garnered 12 million streams a week, holding up over half of its weekly streams at its peak.

Harry Styles - As It Was

In contrast, The Weeknd’s “Sacrifice” got almost 9 million streams in the week it debuted on Spotify, but users played it less than 5 million times the week after it debuted. Not surprisingly, Integr8 New Music Research clients quickly saw their listeners had little passion for this song.

The Weeknd - Sacrifice

2) Ignore rank in streaming data. “As it Was” and “Anti-Hero” both debuted as the week’s #1 most streamed song on Spotify. However, so did the Weeknd’s “Sacrifice”, even though it got less than one-third as many plays the week it debuted. A big debut by a major artist such as Drake or Taylor Swift will knock hit songs out of streaming charts’ Top 10, even though those songs saw little erosion in actual plays. Bottom line; look at the raw number of times streaming users play a song from week to week, not changes in chart rank.

3) Look for songs that are growing week to week. The norm for streaming data is to see big debut weeks. Not all songs work that way. As observed earlier, Lizzo’s “About Damn Time” grew in its first few weeks. As the rare artist whose core fans are older than her casual fans, Lizzo’s “About Damn Time” didn’t see a burst of streaming activity as fans flocked to hear the song right away. Instead, causal fans likely discovered the song on the radio, then went to Spotify to hear it again.

Another example is 2022’s surprise hit “We Don’t Talk About Bruno.” Neither the song nor the movie Encanto were hits when first released in late 2021. But in January 2022, a cold snap and a COVID wave left parents working at home with their kids. Encanto not only came to their rescue to entertain the homebound kids, parents discovered they liked Bruno even after the kids were in bed.

In short, “We Don’t Talk About Bruno” was a song that picked up a much wider fan base than the core Disney fans who played the song when the movie debuted.

Encanto Cast - We Don't Talk About Bruno

Let’s apply these concepts to a song that—as of this writing—is relatively new.

Miley Cyrus - Flowers

After peaking with almost 22 million streams the week after debuting, Miley Cyrus’ “Flowers” maintained 85% of its weekly plays on Spotify in the U.S. the following week. This pattern suggests—and Integr8 New Music Research clients can confirm—that people are playing “Flowers” because they really like the song, not merely because they’re curious what Miley is up to.


I’ve been urging our clients to use streaming data as one of their new music intelligence tools ever since they were still tracking CD sales. Sadly, many programmers are still hesitant to trust streaming data because its metrics are unfamiliar and confusing. It is my hope that by empowering programmers to understand the story each song’s weekly play counts is telling about each song and its artist, radio programmers can confidently maximize this amazing resource.

However, streaming plays can never tell you everything you need to know about a song to know if your station should play it. It can’t tell you how many listeners know each song, which segment of your audience is most passionate about each song, how many listeners hate a song enough to change stations if you play it, and when listeners are ready for you to move a song to recurrent rotation. For these essential data, high-quality new music research remains the gold standard for selecting songs best aligned with your music strategy.

Want to go deeper? Join Integr8 Research President Matt Bailey for Retention Rate: The Magic New Metric for Measuring Hit Potential in Streaming Data: The Webinar Tuesday, April 19th at 2PM EDT/11 AM PDT.

Registration is now open here.

YouTube Music and Spotify on the Podcast Branding Carousel

As I reflect on the past week in Las Vegas at the Podcast Movement Evolutions conference, I’m thinking about two recent announcements that will have a significant impact on the podcasting industry. The first announcement, made the week prior at the Hot Pod Summit in New York, was revealed by YouTube’s head of podcasting, Kai Chuk.

YouTube will begin featuring audio and video-first podcasts on its YouTube Music platform.

The level of YouTube’s prowess for podcast consumption and discovery and this move was widely buzzed about in Vegas. For many podcasters, how to utilize YouTube effectively is a conundrum. Unlike audio-only platforms like Spotify and Apple Podcasts, YouTube cannot pull your show in with an RSS feed, which automatically populates artwork and show notes, and results in the user-friendly displays and players you see on audio platforms. Therefore, podcasting on YouTube can be a clunky, manual process. Discovery can be difficult and sorting chaotic. Because there’s no RSS feed, podcast analytics are a challenge, and thus monetizing becomes a pain.

Within this context, YouTube’s move makes sense. By offering podcasts on YouTube Music, it will pull in RSS feeds and should be easier for podcasters to set videos as podcasts on YouTube Studio. But YouTube’s biggest challenge has nothing to do with easing functionality for podcasters and listeners.

It faces a massive branding challenge.

According to Viralyft, YouTube has over 2.68 billion active users globally as of September 2023. YouTube Music just surpassed 80 million subscribers. YouTube is where podcast listeners are. YouTube Music is where they want them to be. In theory, it will create more paid subscribers for YouTube Music and it will offer an ad-supported version if you don’t want to pay for it.

But YouTube Music is not a podcasting platform today. It never has been. It has “music” in the name and will attempt to grow using a spoken-word medium. It will take an immense effort to educate consumers of YouTube Music’s new role as a podcasting platform, one with no guarantee of succeeding. It may be challenging to improve the podcast experience on YouTube, but that feels like the more logical branding play. YouTube Music as a podcasting platform will start as a weak brand in the podcasting space, and we won’t know the quality of the content until its planned launch.

Anchor, one of the top podcast hosting platforms, was acquired by Spotify in 2019 and it benefitted from the unique new show boom that was fueled during the pandemic while people were at home. The number of active podcasts inflated. One of the things that supercharged Anchor’s growth was the fact that it was, and remains, a free hosting service and attracted beginners. This is in contrast to the $15 or so monthly fee that most hosting platforms charge to house your podcast, distribute to multiple players like Spotify and Apple Podcasts, and provide a level of analytics.

Then, all of a sudden last Wednesday at Spotify’s Stream On event, the company announced it would change the name of Anchor to Spotify for Podcasters. If you do a Google search for Anchor podcasts, you may see Anchor in the listing but you’re redirected to podcasters.spotify.com with a long explanation of why Spotify for Podcasters will be better. Among those improvements include the ability to upload video podcasts to Spotify and more advanced analytics.

Just as time will tell if YouTube’s launching of podcasts on YouTube Music will work, we’ll have to see how Spotify’s dropping of the Anchor (sorry, too easy) plays out. The company didn’t spend time leading up to the change educating and informing consumers it was coming. Many users will be confused when greeted with the new name. Spotify is a podcasting player, not a hosting platform for podcasters. Its challenge will be to convince current users to stay, as well as educate consumers about Spotify’s new role as a hosting platform.

In time, both initiatives may succeed but each will be undeniably difficult. Changing brand perceptions to influence consumer behavior is one of the most difficult tasks a marketer will face. We look forward to watching both closely.

Four Questions to Answer in Audio Brand Perceptual Research

Tuesdays With Coleman, our blog that offers tips and insights on branding, content, and research strategy, is now four years old. Over the next four weeks, we’ll reprint four blogs (one per year) that made the most impact through industry engagement.

This week, we’ll start our celebration of the number four by presenting four of the most important questions that can be answered in perceptual research–a type of study we conduct often at Coleman Insights.


Whether it’s a radio station program director poring over ratings, a podcaster dissecting downloads, or a streaming service examining the subscribers of one of its channels, there is a big question that these numbers don’t answer. When consumers choose an audio brand to listen to, is yours one of the ones they think of?

The measurement we use to obtain this information is called Unaided Awareness, and there’s a reason it is one of the first questions we ask in perceptual research. The goal is to learn the first brands that come to mind in your listening universe. For a radio station conducting a perceptual study, for example, we may ask respondents to name as many radio stations in their area as they can remember, regardless of whether they listen to them.

When it’s time to pick an audio brand to listen to, your target consumer is only thinking of three or four at any given time. How can you generate significant listening numbers if your brand isn’t one of those top-of-mind few? It’s no different from the exercise your brain goes through when you pick a restaurant to take a friend to lunch. If the restaurant isn’t top-of-mind, you’re probably not eating there. Radio is at a further disadvantage here, because while a consumer may use a tool like Yelp or Google to discover a restaurant or Apple or Spotify to discover a podcast, radio discovery tends to be more organic or reliant on paid marketing.


In August, Sam Milkman’s Tuesdays With Coleman blog “I Can Tell You How Healthy Your Brand Is With One Question” took a deeper dive into the “first thing that comes to mind” philosophy. Not only should your brand be top of mind, but your listeners should also be able to explain what your brand represents in a few words. “That station plays hit music.” “That station plays New Country.” “That station plays Hip Hop.” “That podcast is about serial killers.”

It seems simple, but you may be surprised how often listeners are either not able to answer the question at all or think of your brand for other things first. For example, listeners may think of your station for playing the most commercials or having too many contests before they think of it for its music or talk position. It’s a challenging but correctable problem, but research can pinpoint where the issue is.


For music radio stations, there is often no more important question than this but there are many ways to approach it. 1) Are you playing the most popular music? Not just with your current listeners; are you playing music that’s popular enough to attract new listening or have you maxed out available audience with the styles you currently play? 2) Are you getting credit for the right music? Are listeners thinking of your station for the music you play or is another station getting credit? Are they thinking of you for styles you don’t want to be associated with? 3) Do the music styles I’m playing work together? The reason why Pandora picks the next song for you and Spotify curates personalized daily mixes are not by chance. They are highly data-driven algorithms based on your music preferences. Radio stations can have a similar advantage by using Compatibility data to learn which styles are more likely to create tune-ins and tune-outs when played together.


Just as it is important to have a baseline of Unaided Awareness to learn how many consumers are thinking about your brand, it is also helpful to know how familiar your key personalities are in the market. In a Coleman Insights Plan Developer study, respondents only evaluate a personality if they have heard of them. Thus, you can see which personalities may not be very well-known but show a great deal of upside with a positive evaluation. Or vice versa, a personality that may have challenging evaluations that need to be addressed and coached.

While every perceptual study is customized based on the issues and challenges germane to each specific brand, these four important questions are the backbone of a great many of them and provide the stepping stones to actionable strategic plans.

From all of us at Coleman Insights, have a very Happy Thanksgiving. Next week we’ll begin sharing the most impactful blogs from the past four years – starting with 2018.

How Platform Choice Impacts Contemporary Music Tastes

Tuesdays With ColemanOver the last couple of weeks, we’ve learned quite a bit about the current state of contemporary music. Among many other findings, this year’s study of the current tastes of 1,000 12- to 54-year-olds across the United States and Canada has indicated a rise in the appeal of Country, a slightly older lean to the best-testing titles and a downtrend for Pop, Hip Hop/R&B and Dance/Electronic. This week, we’ll focus on how the genres of the best-testing songs vary based upon consumers’ choice of platform. For example, the best testing genres among radio users look different than those of streaming users. Pandora fans look different than those consumers who prefer Spotify. Why do we find these differences so interesting? Because programmers are barraged with data from different sources every day. A song’s amazing number of streams on Spotify, for instance, might be used as an argument why it belongs on your radio station. Or the fact that “everyone” on Pandora is flocking to a particular style suggests that you should move your programming in that direction.

But is it really that simple?

The Weeknd’s “Blinding Lights”, which has strong radio airplay, is #1 on the Spotify Top 200 Chart. “D4L” by Future, Drake & Young Thug debuts at #1 on Pandora’s Top Spins chart. “The Scotts” by The Scotts (Travis Scott and Kid Cudi) which debuted during a live virtual Fortnite event that attracted over 27 million unique participants, bowed at #1 on the Billboard streaming chart last week.  Does that mean these are the most popular songs in North America—or that they are popular among people who listen to your format? Not necessarily, particularly if the people who are using radio or streaming on a daily basis have different music tastes.

That’s why understanding the different profiles of consumers of these various platforms should matter to you. It should help you appreciate what all those stats being thrown at you really mean.

For starters, the best testing songs of people who use radio every day look a lot different than those of daily streamers. What’s the big difference? The Top 100 among daily radio listeners contains a large percentage of Pop and Country, and a smaller amount of Hip Hop/R&B. About a third (32%) of the Top 100 of Daily Radio Listeners is Pop and 29% Country, but only 19% Hip Hop/R&B. Daily Streaming Listeners, on the other hand, have much more Hip Hop/R&B (29%) and far less Country (only 15%).

Does that mean Daily Radio Listeners don’t like contemporary Hip Hop? No. It means when we look at Daily Radio Listeners as a group overall, they gravitate toward Pop and Country among contemporary genres. You are more likely to find interest in Pop or Country when you take a broad look at regular radio users.

We see other notable differences when we compare the Top 100 of Pandora, Spotify and YouTube fans. Consumers who prefer Pandora over other streaming services have a tremendous amount of Country in their Top 100—39%. They also have 26% Pop but significantly less Hip Hop/R&B at only 17%. Those who prefer Spotify go in the opposite direction. They have a very large percentage of Pop (39%) and a good amount of Hip Hop/R&B (26%)—but very little Country, only 9%. YouTube fans look very similar to Spotify fans.

The point is that people who prefer Pandora have much more Country in the songs they rate best; those who prefer Spotify and YouTube have more Pop and Hip Hop/R&B in their Top 100 songs. We sometimes tend to think of streaming users as homogeneous, but they are not. The profile of consumers who prefer different streaming services are distinct—and it is important to keep this in mind when we look at data coming from various sources. And that’s true of almost every different platform we analyzed.

Next week, we’ll dive into the political fray–to discover the respective taste differences between supporters of President Trump and Joe Biden. In an environment in which common ground and bipartisanship can be hard to find, can these two polarized groups find musical consensus?

Don’t miss next week’s Tuesdays With Coleman to find out.