The Fields of Dreams principle — “if you build it, they will come” — doesn’t apply to audience attention anymore. It’s not enough to just create media; you must also attract an audience to consume it. Capturing public attention now requires both making media and making audiences.
In his new book, The Marketplace of Attention, James G. Webster, a communication studies professor at Northwestern University, explains how audiences form in the digital age and the factors that influence them, including user preferences, social networks, as well as ratings and recommendation systems.
Webster argues that while the digital age empowers individuals to choose where and when they consume media, there are forces like algorithms and social networks that shape those decisions — sometimes in unnoticed ways. These can both fragment and concentrate audiences. In fact, “the forces that will concentrate public attention are underappreciated,” Webster writes in the book.
“[Publishers] certainly have to be mindful of what people are interested in and their preferences, but realize that their preferences are variable and that sometimes even when they know what they want, they’re not particularly good at finding what they want,” Webster tells me. In other words, publishers have to continue creating compelling content, and also better understand their audiences and harness some of these forces.
I talked to Webster about the model of “massively overlapping audiences,” the subtle ways preferences are influenced online, how the “time spent” metric is misunderstood, and the importance of a publication’s brand and editorial judgement in fostering loyalty.
Even when people know what they want, they’re not particularly good at finding it.
Let’s start with your assertion that almost every media outlet shares an audience with every other outlet. In other words, no outlet lays exclusive claim to an enclave of loyalists. Can you expand on this?
As we move into an environment where people have abundant choice — content of all different kinds, liberal, conservative, entertainment of every imaginable sort — there is a tendency to say that individual media users now have free rein to consume whatever they want, wherever they want, whenever they want. In that kind of environment it’s tempting to assume that if they like a particular kind of programming or a particular ideology, that’s all that they’ll consume — that it’s the only thing that they’ll be exposed to.
But when you actually look at patterns of audience behavior to see if you can behaviorally identify these camps of loyalists, they’re very very hard to find. What you do find and what I talk about in the book is this tendency for audiences to overlap — that the people who visit one extreme right wing site are also quite likely to visit other mainstream sites.
Then the question is, why aren’t people behaving as theory, or even to some extent common sense, would suggest they ought to behave when they have all of these choices?
Can you explain some of these forces that create overlapping audiences?
One is about the users themselves and just how well-defined are their preferences. For most people, most of the time, their preferences aren’t nearly as crystal clear as theory or common sense would have you believe. Quite often people are much more omnivorous than we sometimes give them credit for being. So that’s a part of what’s going on on the user side.
You would think that this superabundance would facilitate freedom of choice, but people have to develop strategies for coping with all of that abundance. What they typically do is narrow their choices to a relatively small repertoire of websites, channels or outlets so that they don’t have to search the entire World Wide Web. They also use rules of thumb called heuristics to make quick judgements about what would or would not be of interest.
When you actually look at how people behave, they don’t adhere to simple-minded genre or ideological preferences.
Part of the problem is that media products typically are “experience goods,” so even if you’re familiar with a particular television program, you don’t know until you’ve experienced the latest episode whether that one is going to deliver the goods. It’s only after you’ve consumed something that you know that it is what you wanted, or not. So there are a lot of reasons why people’s media choices are not completely rational or don’t look rational, even if they’ve got well defined preferences.
You run into other problems when you begin to consider structures. Structures of the larger media environment have ways of making some things easier to find, encouraging consumption, or making other things drop out of view. The result of all this is that when you actually look at how people behave, they don’t adhere to simple-minded genre or ideological preferences.
A lot of what you’ve said about people being omnivorous largely comports with a survey we released in March about how Americans consume news. I’m curious about what you call “structure driven loyalties.” You argue that what can appear to be loyalty to a particular outlet or genre can actually be driven by the structures of everyday life. What do you mean by this?
If you’re a media outlet, it would perfectly logical for you to define your loyalists as the people who are spending the most time watching your channel or website — people who are the heavy users of what you’re offering in the marketplace.
Almost always, the people who consume a lot of a particular channel, the heavy users, are heavy users of the medium in general.
Think about public television, for example. I used to work in public television years and years ago. I think there’s a common notion among public broadcasters that their audiences may not be very large but they’re loyal — that they don’t watch a lot of television but when they do, they watch public broadcasting. Turns out, that’s not the case. Almost always, the people who consume a lot of a particular channel, the heavy users, are heavy users of the medium in general. So for an awful lot of outlets, you see a small percentage of users accounting for a disproportionate amount of the audience for that outlet or the time spent with that outlet.
It sort of comports with the 80-20 rule you run into in marketing, where 20 percent of your audience is doing 80 percent of your viewing. That’s true in public broadcasting. But when you look at those 20 percent and ask if they are only watching public broadcasting, the answer is no — they’re actually spending only a small percentage of their total television viewing time with public broadcasting and they’re watching a whole bunch of other stuff. They, in particular, are all over the map. They just watch a lot of public TV because they watch a lot of TV. I don’t have evidence of other media besides television, but I would bet that’s true in most other instances.
Does that same logic apply online? If you’re a heavy internet user, you’re more likely to read more news sites?
During API’s summer meeting in Chicago on metrics and measurement for news, Pete Davies from Medium had an extended argument about the utility of attention and time spent. It was specifically time spent and I thought that that was probably a more useful metric than page views or unique visitors. But what he’s probably not thinking about is that most of the time spent viewing on his site is probably generated by a very small number of readers of his total readership.
So as an advertising metric, if I was a cynical advertising person, I would say, well it’s easy to get people who spend a lot of time reading online, they’re not scarce. So why would I go after people who are spending a lot of time online? They’re probably heavy internet users and I can buy them easily with programmatic buying because they’re all over the internet.
“Ultimately, the most important thing to know is what else those readers are doing and that’s harder to see.”
I thought that compared to the other common metrics that were being talked about, that time spent probably did a better job of tapping into some level of engagement. I was just trying to think, well, is there some fly in the ointment with going with time spent? If his readers are this small but loyal audience of people that go to Medium and spend a lot of time, then maybe they’re not anywhere else on the internet and that would be extra valuable for advertisers to know about. He can see what’s happening with visitors spending time on the site, but ultimately, the most important thing to know is what else those readers are doing and that’s harder to see.
So everyone has to build these audiences. You mentioned that there are two methods of doing this: push and pull. Very simply, push is when the media seek out the users and pull is when users seek out media. You write that, “frankly, the instances of genuine pull audience building are few and far between.” What do you mean by that and who’s doing it best?
One of the reasons that I addressed the issue of pull and push is that you run into this sentiment — and it sort of goes back to what we talked about initially — that in a world with anywhere, anytime media consumption, people are liberated to pull whatever they want. And along with that is you get people who sing the demise of push media.
There have been a few people, like Eli Pariser, who have written about it but when you start thinking about it, the instances of 100 percent pull are really hard to think of. A true pull media is a sort of like what I refer to as the “field of dreams” strategy, build it and they will come — where you don’t do anything to push it out. You don’t publicize it. You don’t market it. You don’t do anything.
As an exercise, how would that happen where you can just create media, not do anything and you’d draw in an audience? There might be instances of some innocent video, like “Charlie bit my finger,” where you post it and social networks just latch onto it and it becomes this huge viral thing. But even with that, you’ve got to ask how did all of those millions of people come to watch the video? And they were probably got nudged by a friend, a link, this or that, or they saw it covered as a human interest story on good old fashioned mass media and then they went and found it.
You don’t think of them as heavy handed pushes but there are all of these forces that tend to privilege certain things and draw attention to them and invite people to follow the link. It has largely to do with the structure of media. It also has a lot to do with media metrics and the recommender systems that encourage these viral explosions.
You write that good media filters cultivate, rather than simply cater to preferences. How can we build a better filter?
Well there are a lot of things going on there. We’re going to have to unpack the question a little bit.
If you spent your entire life on Facebook and were subject only to the EdgeRank algorithm in your news feed, then that would probably constitute a closed system. I’m hoping that most people who use Facebook don’t spend their entire lives on Facebook.
This goes back to the observation that Facebook — for all its popularity and success — is not an enclave. People who use Facebook use a bunch of other websites and they probably watch television and read other things online, so they’re all over the place.
The argument that we’re completely at the mercy of unseen recommenders and algorithms and so forth is most completely developed in Eli Pariser’s The Filter Bubble, which is chilling but it can only work to greatest effect if each one of us lives in a bubble that we never leave.
My argument is that that doesn’t happen.
You’re subject to a bunch of cross currents and — maybe this is wishful thinking on my part — that with the proliferation of platforms and all of these choices, it’s harder than ever to corral people into one bubble where you control everything that’s brought to their attention. So that’s part of the answer.
The other part of your question had to do with cultivating preferences. The standard assumption in almost all social sciences is that people bring their preferences to the media system and that they act upon those preferences. So preferences drive your choices.
So they consider the possibility of exogenous preferences, which are from outside the system. They rarely consider the possibility of endogenous preferences, that by living in this media system it cultivates certain preferences that you wouldn’t otherwise have. To me, that’s almost certainly the case.
Finally, the last part of your question was about building recommender systems and cultivating preferences. Changing preferences with recommender systems might or might not be a good thing for the reasons we talked about. What people who think about say is, what if we didn’t always give people what they were predisposed to want? What if we could build a little bit of serendipity into recommendations? Wouldn’t that be a good thing?
And there you’re kind of in a dilemma. The big picture is a little bit of serendipity is probably a good thing so we don’t default to what’s most comfortable and we’re exposed to different ideas and so forth. But pragmatically, if I was using a search engine or a recommender system that fed me a lot of serendipity my reaction would be it’s not giving me what I want. It’s supposed to make my life easier, not force me to weave through these wildcards.
You have to strike a very delicate balance between just a little serendipity so we don’t all devolve into these echo chambers or enclaves, but you still have to be pretty responsive to giving people the things that they’re after, the things that they would like, the things that would be useful at some point in time. So it’s tricky.
The old view of audience is that publishers see people as members of one mass audience. The newer way is to see users as members of one or more networks who are aware of each other and may affect each other’s behaviors. You say that today’s audiences are a blend of the two. How can news providers use this insight?
With most forms of mass media, the traditional way to think of audiences was as a mass audience. The formal definition of a mass, from sociology, is that it’s composed of individuals who are basically anonymous and act independently.
The big sea change in digital media, particularly with the advent of social networks, is that people are less and less anonymous and they can act in concert. They can affect each other’s behaviors — this clearly happens when something goes viral or simply gets widely distributed within some particular network.
Many people have said that a pure mass audience was always sort of an abstraction and people were always talking with one another, but with the digital networks that we have now — it’s interaction on steroids. It’s beyond just interpersonal. So the existence of social networks is not new but they’re now a much more formidable phenomena or force to be reckoned with.
Many people have said that a pure mass audience was always sort of an abstraction and people were always talking with one another, but with the digital networks that we have now — it’s interaction on steroids.
There are certainly people who have looked at what makes content go viral, and it’s the things that you would expect — the things that are outrageous or awe-inspiring or angering. But it’s really hard to guarantee that something will be shared or go viral just by the content you put into it.
There is research done by Duncan Watts, principal researcher at Microsoft, among others that most things just don’t go viral, they just don’t catch for whatever reason. So between not knowing the magic recipe for content and the sort of unpredictable nature of social networks and who is actually going to share, it becomes very hard to prescribe to an editor or a journalist to “do this and you’ll multiply your audience a hundred times over.” It happens. We know it happens, that some things, some stories, some images, some kitty videos just take off, but knowing the recipe for doing that is another matter.
My questions so far have just been triangulating on one big question — how do you build an audience?
Actually in the preface of the book, I try to discourage readers from thinking I’m going to give them the magic recipe. The best that I can do is to discourage people from opting in to sort of simple-minded, if pervasive, myths about how audiences take shape in the digital environment.
I’ve talked about some of these like those myths about consumer sovereignty and preferences. A more realistic way to look at it — the sort of big picture — is to say that you certainly have to be mindful of what people are interested in and their preferences, but realize that their preferences are variable and that sometimes even when they know what they want, they’re not particularly good at finding what they want. So users aren’t nearly as rational as we might imagine them to be.
The best that I can do is to discourage people from opting in to sort of simple-minded, if pervasive, myths about how audiences take shape in the digital environment.
The media environment itself is way more active than people realize. Pragmatically, if you’re a media maker, then you’ve got to think, “How can I harness some of these forces to my advantage?”
This isn’t secret, people have figured out that it might be worthwhile to do search engine optimization, or they’re clever about exploiting social networks or social media. It’s not brand new stuff. But I think in the old media environment it was a lot easier to tell what was going on. If you were a big legacy paper or broadcaster you had an advantage in that what you put on primetime television or on the front page of a big daily newspaper didn’t have to work too hard to find the audience. Now there’s so much more competition that you just have to be more mindful of all of these forces that shape whether or not people will pay attention to what you’re offering.
Ok, one last question. What’s the significance and role of editors and that editorial sensibility — that human algorithm so to speak?
Good question. Even though I point out the idea of massively overlapping audiences and cultures and so forth, it’s still certainly possible and desirable to cultivate audience loyalties. In part because people are in some ways more dependent upon their tried-and-true repertories than ever before. One way to create that kind of loyalty is through crafting a product that people find useful, enjoyable, and all those things you would want it to be.
I think that invites you to think about maybe what I refer to broadly as “preference-driven loyalties.” All you have to do is create something that the right people will like and they’ll just come back again and again. I think while you’re doing that, you also have to be mindful of these structural forces that we’ve been talking about, though. So it’s not, “if only I make my editorial content better,” or “get just the right mix” that’s all you have to worry about. You also have to worry about things outside of the editorial content of the paper or the outlet or whatever medium you’re talking about.