I was on Facebook earlier today and came across the comment below from Tim Quirk. It’s apparently about Apple’s new streaming music service, which is slated to launch later this year. Tim’s a musician and has also spent time working at Rhapsody and Google Music, amongst other places.
WHO’s telling you to listen is far, far less important than WHAT they tell you to listen to. Also, GETTING you to actually listen.
Reading the comments underneath Tim’s post, I came to understand that it related to his criticism of what he asserted was the personality-based approach that Apple seems to be using with its new service.
One commenter, Jamie Dolling, asserted that the new service was as much about “personality as it is about music,” and worried that this approach couldn’t help but “poison the water.”
Subsequently, Jon Maples, a digital music consultant formerly of Rhapsody, indicated that it was the same..old…’human music curation’ approach without any understanding of what the listener actually wants.”
I started adding my own comment on that post, but as its length ballooned it seemed to be morphing into a blog post. So I moved over here.
Anyway, here’s my take on this stuff.
As Tim correctly points out, getting listeners to break out of their default is the challenge, because most listeners just want to keep listening to the stuff they are already familiar with (and consume the same products over and over again as well).
To my mind, personality-based approaches to marketing and publicity can be one way to accomplish that goal. Even if there’s an element of bullshit, snake oil to it, they wouldn’t still be using it if personality-based publicity and marketing didn’t sell shit.
For many people, who is telling you to listen/buy is inextricably bound up in what they are telling you to consume. That’s the whole point. It’s a gestalt experience.
The further you move to the right in the diffusion curve on anything new, the less likely consumers are to use their own research and judgment when making buying decisions. These people are much more likely to look to trusted people (opinion leaders) for signals about what choices to make.
This is not a new thing. It’s as old as consumer capitalism (or actually even older still). And while I pride myself on doing my own research on lots of stuff and making my own decisions, there’s lots of other stuff where I just don’t care enough to do that. I want somebody else to point me to an answer that is good enough (or better still, great). The whole point of doing that is that I don’t know what good is, so I can’t really effectively evaluate the quality of the thing being recommended. If I could, I wouldn’t need somebody else’s advice.
So the personality/trustworthiness of the person making the recommendation is extremely important. It becomes a proxy for the knowledge I lack about the thing being recommended, because, rightly or wrongly, I do feel comfortable evaluating what I think about the person doing the recommending.
The logic goes something like this. I don’t trust my judgment about what cool clothes are. This person over there seems to be a person who I think is cool and has on cool clothes (or at least a person who I understand from their reputation is supposed to be cool and wearing cool clothes). Therefore, their judgment about cool clothes is probably better than mine. So I’ll see what they think I should wear, and moving forward I’ll look to them for more clues and cues on that subject.
Unlike data driven metrics, this isn’t just about figuring out what I want. It can also be about creating a new want, because while I may know what I think I want, I may not actually know what I want all the time. Data driven metrics may do a good job of figuring out what I think I want right now or even what I might want based on what I have wanted in the past, but they don’t do such a good job of determining what I don’t know that I want right now but what I might nevertheless want in the future if it was put in front of me in the right way. The right sort of charismatic curator/opinion leader has the ability to do that.
For many people, music is something they are unsure about. It’s also something where they don’t want to have to filter through all the noise to get to the signal. They like having music around. They also know that what music they like says something about them, so there’s something at stake there beyond just the hedonistic experience of consuming the music.
But in many cases, they just want to be pointed towards some good stuff. If liking this good stuff also seems to help make them seem a little bit less uncool, well, even better. Because in 2015, nobody wants to seem uncool, not even middle-aged people like me. Indeed, seeming/feeling less uncool may be just as important–or even more important–than liking what has been recommended.
So yes, Apple’s new music service is a publicity/marketing platform. And yes it appears to be personality-based. That’s because the biggest objection that the music industry seems to have about many of the other streaming platforms is this:
They may deliver a good user experience to certain users, but despite many assertions to the contrary, they have not yet proven themselves to be particularly good marketing/publicity platforms for companies trying to focus demand on a limited slate of new releases (the only way to generate the kind of cash they need to stay in business long-term). They can service demand when it arises. But they don’t drive demand or significantly shape it.
Moreover, to the extent that these services create new wants in people, the want pattern is much more diffuse than in the old system. So the old-line music industry is still trying to find a marketing/publicity platform that looks and works more like terrestrial radio did in the glory days, because that was a great platform for focusing demand on a limited slate of new releases. It had a focused want pattern.
It’s a fair criticism to say that trying to find that sort of platform is a pipe dream. That we’re in a new reality now, and the desire for that sort of platform reflects an unwillingness to get with the times. But there remain very practical reasons why that sort of platform would still be useful to the music business. So it makes a certain amount of sense that its members continue to chase it.
Apple seems to be taking a stab at trying to provide that sort of platform with a more personality-based approach. But just because Apple may, in part, be trying to solve a problem for the music industry, that doesn’t mean their solution is inherently at odds with the user/audience.
After all, terrestrial radio has often been a personality-based marketing/publicity platform both for labels and all the advertisers that subsidize it. But it’s also beloved by many users, because they value both the curation and the personalities they find there. Often, those things cannot be separated.
That doesn’t mean that human curation is always good. Indeed, on average, algorithmic approaches may now be better at delivering a good-enough experience that is more personalized than the average human curation experience.
But when human curation is good, I think it remains the gold standard for curation, even when it is less personalized. Maybe I’m showing my middle-age here, but that’s how it seems to me. That sort of curation is inherently personality-based. That’s a big part of its appeal. You trust the curator enough to give up control and let them take you on a journey of discovery.
In the process, you bond with them, for being associated with a cool personality has the capacity to make you feel cooler and a part of the world they have created around their personality. That experience creates a want in you.
An algorithm rarely makes you feel cooler like this, because it’s a tool. You might use it for the purpose of doing your own research and discovery. It might even show you some things about yourself that neither you nor other people readily see. That, in turn, might allow you to feel cooler when you deal with other people, because of the knowledge you’ve gained. But even when the algorithm is doing a good job delivering quality suggestions to you, it still makes you feel a little bit more like a data point and less like a human.
A friend of mine recently started a Spotify mix-tape group on FB. Each week a different member delivers a 90 minute playlist to the group (a virtual mix tape). So far, this experience has been infinitely better than any algorithmic experience I’ve ever had, because each group member actually takes into account what other people have done and who the audience is.
So if somebody included a track last week, that track isn’t likely to be in this week’s mix and more than likely neither would that artist. Although if it made aesthetic sense in the context of the mix to include the same artist or track two weeks in a row, maybe it would be in there anyway. But in any case, these mixes have a much richer sense of the many contextual factors that contribute to creating a good mix. The same is true of a great show on a non-comm radio station like KEXP. As a result of this, these kinds of mixes reinforce a sense that the group members are part of something bigger than themselves.
Of course, if that mix tape group was me and 25 kids who are under 15, the quality of the curation probably wouldn’t seem as good to me, although I’d probably still hear an occasional great tune I would have missed otherwise. I’d also feel more like an interloper in that group. Maybe that distinction is actually demographic rather than personality-based. But to me, issues of demographics and psychographics are embedded in the idea of personality-based branding. You are buying the gestalt experience that you associate with that person or company.
This is why an anonymous human curator is less valuable than a curator with a personality/reputation that is known and trusted by users, even if the choices of the anonymous curator are objectively just as good as or better than those of the known curator. The lack of an identifiable personality makes it harder to evaluate the utility of the suggestions. And once you’re dealing with that problem, you’re pretty much right back where you started. The curator is no longer solving a problem for you. Now, you need a curator to sort out the anonymous curators for you.
Don’t get me wrong, it’s not that I don’t use algorithmic radio ever. I do. A group of my college friends made a playlist that encompasses many of the songs that were on the jukebox in the coffee house that was in the basement of our dorm at the University of Michigan (the Halfway Inn). Spotify generated a radio station based on that playlist, and it works pretty well, although it still does a poor job of managing repeats of the same song and multiple related songs by the same artist or by related artists (e.g,. Velvet Underground and solo Lou Reed).
A good human curator does not do these things. That’s part of the artistry. They take those things into account. Like I said above, they have a better and richer contextual awareness. Also, part of the reason that particular Spotify station works as well as it does is because it’s based on a playlist that was human curated. So it’s bootstrapping on the contextual awareness of the people who compiled that list.
If enough people trust a guy like Zane Lowe, some of that is his personality. But his personality and that trust is also a function of his talent for curation.
Grunge/Alternative rock broke on commercial radio back in the ’90s in no small part because of Seattle DJ Marco Collins. The relationship that people had with Marco at that time was very much personality driven. They liked him. He was a dynamic on-air personality, and they thought he was cool. But a big part of the reason why was that they came to trust his taste.
If Marco said something was cool and played it on the radio, people gave it the extra listens they needed to appreciate why it was worthwhile, even when their first impression might not have been great. Twenty years later, his impact is clear enough that somebody recently made a movie about him.
Some people have that intuitive gift for knowing what new stuff people will like if they just give it a chance. Computers are also getting better and better at deducing that information based on prior user behavior. But I’m still not sure those two approaches always lead you to the same place.
What’s that term? “Filter Bubble,” where your perceived options keep getting smaller and smaller as the search algorithm feeds back based on your previous choices. At its best, human curation seems less prone to the filter bubble (although it has its own problems and risks–e.g., it’s probably more prone to personal politics and lobbying, which can create a bureaucratic capture problem that undermines trust–See e.g., payola). But human curation only works if people trust the human curators and don’t have to invest too much energy vetting them.
Apple is a high profit-margin, gold standard brand. That’s why people pay extra for it. Its whole message is grace, ease of use, and quality (even if these things are not always actually true). Historically, it’s been about finding the spot where technology and people align. You know, a mix of art and science.
That’s the value proposition it is selling. The personalities are at least theoretically in the service of that. They are supposed to be part of the art that interfaces with the science and tech.
Part of the art is also the fashion sensibility. Undoubtedly, that’s part of what must have attracted Apple to Beats. I have mixed feelings about that. My feathers get ruffled thinking about paying hundreds of dollars for a pair of headphones that may be fashionable but ultimately aren’t very good sounding headphones for the money.
But at the end of the day, I guess I’m a bit of an engineer at heart. I value function over fashion, and I especially hate the idea of paying a premium for something just because it is perceived as being fashionable. Nevertheless, I also recognize that many people are not like that, and that these kind of people are more than willing to pay a premium for something they perceive as fashionable. Indeed, in many cases they are the highest margin consumers.
The personality-based approach also dove-tails with Apple’s history and culture. Before the death of Steve Jobs, it was a personality driven company. It’s also an opinion leader brand. So while it collects plenty of experience data from users, it has not historically solicited explicit input from the public about what it wants. It doesn’t have the same sort of beta-testing developer blogging, two-way conversation that many other companies have as they develop their products.
I once had conversation with a Boeing IT guy in a bar here in Seattle. He said they loved Microsoft, because they were much more open with his department about what they were working on and where it was going.
Typically, Apple hasn’t shared where it’s going until it releases a product for the public to see. It’s not looking for that sort of approval and feedback. When it releases something, the message is this: “Here’s our new thing. We’re cool. We think our thing is cool, and if you try it, we’re confident that you will think our thing is cool too, even if you don’t understand right this minute why it’s cool.”
Over the last decade, it’s had a pretty good track record doing that. So even when it does something that other people have arguably already done, it typically re-contextualizes it in a way that makes it sit differently with the public
We’ll see if Apple’s new music service provides that sort of bold leadership and delivers on the idea that theirs is a place where art does a better job meeting science than at other places. We’ll also have to wait and see whether their approach resonates in the market.
If it just ends up being a re-branding of the Beats music service, then I think the answer will be “no.” While there was nothing really wrong with Beats and it looked cool, at the end of the day, I didn’t find it qualitatively different from its competitors, either in terms of user experience or curation.
So to succeed, imho, Apple will need to extend things considerably on the personality front and keep their curated playlists and other personality-based offerings far more dynamic than they were on the old Beats service. Otherwise, it’s just the same wine in a different bottle.