Where Are We?
Nobody’s Doing Location Right
It’s been 9 years since the first iPhone came out, ushering in the era of location data. In that time, we’ve seen the meteoric rise – and subsequent “plateau of oblivion” – of Foursquare, who vanquished Gowalla. We’ve seen social media flower & fruit, Virtual Reality take root, and, most recently, we’ve seen Pokémon Go grow like an invasive weed ((I think it is de rigeur in think pieces this week to mention Pokémon Go. My apologies.)). And there are now actual experts in interface design, making nearly every mundane detail of one’s life easily controllable from a 5-inch screen; you can fly a seeing robot, lock your front door remotely, or broadcast live video of acceptable (or even exceptional) quality all over the world.
If you tend to be the pragmatic sort, take heart! This progress has not been limited to the mysterious world of revenue-free companies with enormous “user bases” that are growing at astonishing paces. Gaming has hit a whole new jackpot with mobile; worldwide gaming revenues now stand at an estimated $99.6 billion for 2016, of which 37% is mobile (smartphone + tablet). ((https://newzoo.com/insights/articles/global-games-market-reaches-99-6-billion-2016-mobile-generating-37/)). Pokémon Go gives us just a glimpse of the possibilities when you bring gaming content to mobile. Can you imagine that graph as it’s likely to appear in a 2020 article somewhere?
That’s just gaming. Since 2011, mobile advertising revenue has gone from zero to almost $21 billion in the U.S. alone, and grown at a rate of 90% per year ((Source: IAB Internet Ad Revenue Report 2015. . We’re seeing some brilliantly creative expressions of the form – although that’s far more the exception than the rule, in this writer’s opinion.
But here’s the point: all the while, we haven’t seen location data truly mature – certainly not in consumer applications. I have some ideas.
Note: I drafted most of this post just before the Pokémon Go craze. While I think Pokémon Go will be a blip on the radar (much as iFart was in 2008) it does foreshadow substantive change in how location is used in media. I’m hopeful – optimistic, even.
In my admittedly biased view, Pokémon Go strengthens the argument I’m making in this post. One game title added 9 billion dollars of market value to an already iconic worldwide brand — in the first three days. Stories of obsession flooded every channel of the news – giving even the imminent Republican National Convention a run for its publicity money. I was particularly haunted by the grainy, shaky video of hundreds of people gushing into Central Park to capture a virtual creature.
If all that doesn’t point at an inherent and nascent craving for better location-based experiences, I don’t know what does.
Just Media. Why Just Media?
This blog post could become a book without some constraints on our definition of “location data”. (Or even with those constraints, really.) So for our purposes, I will treat location data as it exists in the world around mainstream smartphones (again, including tablets). I will not include automotive GPS systems, avionics, commercial & military-grade technologies, shipboard navigation … or any industries in the “Internet of Things”. I want to talk about location data as it pertains to media.
Smartphones, as we’re all becoming increasingly aware, are far more than just media devices. But despite not being limited to media, smartphones define location-based media:
- unlike any other location processor, smartphones are also true multimedia players
- smartphones, unlike any other location-aware device, can collect data in addition to processing a locator signal. ((before anyone jumps all over me to tell me I’m wrong, I mean data other than lat/long)) This is a hallmark of modern media: the data flows two ways.
- smartphones are, by definition, ultra-portable
So when I talk about location data as it pertains to media, I’m talking about smartphone media.
So What’s “Media”?
Let’s take a quick look at just what media is at its most basic level. In a 2004 consulting engagement for a supplier of consumer marketing data, I defined media as “any business that seeks to monetize eyeballs”. This definition seems no less relevant today.
The “ecosystem” to accomplish this, as one might imagine, was already complex before the age of data and “ad tech”, and has grown exponentially more so. But at the same time, it is an ecosystem that can be explained very simply, because it’s all about 4 main constituents: advertisers, publishers, end users, and sometimes a distributor to connect these:
Publishers, the real engine of this system, have just two choices to “monetize eyeballs”:
- they can get money directly from the end user (i.e., “Circulation” Revenue), or
- they can sell access to those same end users (i.e., “Advertising” Revenue)
Every business model in media is either a derivative or a permutation of this basic theme.
When you define it the way I have, the first true media business was arguably newspaper and magazine publishing. And, indeed, business models from the early days of the Internet were direct copies of the magazine/newspaper model:
REVENUE = Circulation Revenue + Advertising Revenue
REVENUE = (Subscribers x Subscription Price)+(Newsstand Purchases x Cover Price) + Advertising
This core model is the kernel from which all the complex new internet-based business models grew. Every successful modern media business carries that DNA.
In Defense of the Newspaper/Magazine Model
“Old Media” publishers have suffered some humiliating defeats in the past 20 years, and they have missed some enormous opportunities. Magazine & newspaper publishers have been widely analyzed – and derided – for their failure to anticipate and prepare for the fundamental change in content distribution.
But let’s consider one thing that they got enormously right … and long before anyone else did: context.
Contextually Relevant Advertising
So what do I mean by context? It’s not quite a buzzword yet, but it certainly has been getting its fair share of “ink” (pun sorta intended) in the past few years.
As I’m using it, context means the publisher or advertiser knows something about the user and uses that knowledge to match user to content:
The content can be advertising, or “editorial”, or a combination of the two. If it’s contextually relevant, the content both fits together and matches the audience. So you’re not likely to see a full-page ad for asphalt services opposite the cover page for a story in Cosmopolitan magazine entitled “Meet Your Other G-Spot”.
This concept was not lost as content made its transition to online distribution. As soon as online audiences were large enough to support it, data-driven understanding of the audience led to an explosion of advertising technologies like Google’s AdSense, which try to ensure that the end user is well-matched to the message.
Location Data: THE BASICS
For all practical purposes, we can trace location data back to that original iPhone of 2007. Yes, of course, location data existed long before that; after all, the 16,500-year-old star maps of Altair, Deneb, and Vega and of the Pleiades were image data (a drawing) associated with the observer’s location (what is now Lascaux, France).
Actually, let’s look at that example:
At this point, this is data, but it isn’t in a database. It is a placename – Lascaux and an image – which you have to go to the caves to see. But we can easily put it in a database, especially if we have a photo of the actual painting:
|LocationID||Piece of Information|
|Latitude: 45° 2′ 57″ N, Longitude: 1° 10′ 34″ E||ExtPainting.jpg|
This is the basis of all location data. A location identifier – such as a placename or lat/long coordinates – and some information that can be “attached” to that location.
We can define “location” with as much or as little precision as we like; we can attach that image to “EUROPE”, or “FRANCE”, or to the exact 1-meter space where the original image is, or even the 1 cubic centimeter in the exact center of the cow’s eye.
Note that in our example, we are treating location in just two dimensions: latitude and longitude. We have an inherent 2-D bias when it comes to location; this is probably because maps have pretty much always been drawn in 2-D, and this is what we were exposed to from 3rd grade on. But to get to the 1 cubic centimeter in the center of the cow’s eye in the Lascaux cave painting, we’re going to need another dimension: altitude.
Actually, Four Dimensions
Suppose I tell you to meet me at the Bedford Avenue subway station, and suppose that I know you’re not that familiar with Brooklyn, so I even tell you that the subway station is at 40.72° N latitude and 73.96° W longitude. What are the chances we’ll see each other? Well, they’re pretty small, because I left off some critical information. I didn’t tell you if I’d meet you upstairs at the entrance or downstairs at the turnstile. And I also didn’t tell you when.
These are the four dimensions of location: an x-coordinate (e.g., latitude), a y-coordinate (longitude), a z-coordinate (altitude), and a t-coordinate (time). With those four components, we can be very precise with location data.
THE CURRENT STATE OF THE ART
In a story last month (June 2016), Fusion‘s Kashmir Hill shed a little light on location data. The angle: Facebook may be using your location data to suggest “People You May Know”. Facebook, for its part, says it once ran a test but does not use location data as part of your profile. ((You may, of course, choose whether to believe Facebook, which in this writer’s view has a deplorable record when it comes to privacy & data. I personally deleted my Facebook account almost three years ago, but if I still had it, I would most certainly disable Facebook’s access to my phone’s location.))
So how arepublishers and advertisers actually sourcing and using location data today? Let’s break that down.
In the current model, location source data comes down to two essential questions:
- Who is here?
The end user’s smartphone answers this question, using any of the following:
- Cell Tower Triangulation
One quick note on geofencing: a location data feature that draws a figurative fence around a point. This is not a technology in and of itself but is used to determine if a user is in a designated area. So, for example, a publisher may authorize certain “bonus content” for any users who are within a 500-foot radius of a place, or an advertiser may offer a coupon to a shopper who is in a particular aisle in their store.
- What is here?
For example, that prehistoric cave art In Lascaux, France. As we’ve discussed, people have been recording spatial information since long before there were alphabets. Storing this information in modern databases began in the late 1950s/early 1960s.
By 2004, Yelp was adding new Internet-driven “user generated content” capabilities to location data. Both Foursquare and Gowalla launched in 2009, adding a different layer of “what is here?” information to the online databases.
Now there are countless companies, each with its own take on “what is here”, from Waze (“what traffic is here?”) to Placecast (“what deal is here?”) to Yext (“what business is here?”), and so on, fere in infinitum.
Quick: name a location-based app, right off the top of your head.
You said Foursquare, didn’t you? Foursquare may be the best-known company that puts location at the center of its business model. On top of that, they’re an excellent case study in how location is currently used. Foursquare is about what I call Point Content – they tell you about that one place, and generally, they tell you about what’s at that one place now.
You can find out what’s on sale in the Juniors section at Macy’s, or what business is at 315 Bowery. You can find the “best”, or the closest, place for a cup of coffee, or something stronger. You can ask Siri what movies are playing near you, or in your girlfriend’s neighborhood, or in Copenhagen. (All of these, of course, provided that the database itself is up-to-date.) The list really is endless.
“Circulation” and Advertising Revenue
Remember, all of these services are provided by companies who ultimately need to make money, and their presence on the mobile web means that they are participating in a media business. Either they want you to see an ad, or pay for what you’re doing, or some combination of the two.
Since 2007 – again, that’s when the original iPhone came out – location data has gotten more sophisticated, and it’s certainly gotten more voluminous. Location data has become increasingly incorporated into our smartphone-driven lives. There are location DSP s, DMP s, and SSP s – “programmatic” specifically for location-based publishing. These algorithm-intense systems are designed to optimize the Advertising part of the equation.
And modern Content Management Systems (CMS) are similarly designed to optimize the content ⟷ end user relationship, which can not only effect content pricing decisions, but as a side benefit can also effect advertising rates.
But it’s all still point content.
This treatment of the current use of location data would be incomplete without some discussion of analytics. Because, yes, location data is definitely being analyzed.
I mentioned earlier in this post that the hallmark of modern media is that data flows both ways. In the old days of media – the days of broadcast television and printed magazines – the data (that is, the content) only flowed from publisher/advertiser to end user. (If you’re interested, here’s what I said in 2004 about how “media is different”.)
In what may be the most ballyhooed example of hard-core analysis of location data, Foursquare used foot traffic around Apple retail stores to predict opening-day sales of the iPhone 6. And they pretty much nailed it.
I did not consider analysis a third revenue stream in media, because fundamentally, it really is a derivative of advertising. It is the sale of bundled eyeballs.
So What’s Wrong?
1. Advertising, in general, is broken
Let’s get this out of the way. Advertising is broken and has been for some time. We all know this. Every time we see a Toyota commercial, which is indistinguishable from a Ford commercial, which is indistinguishable from a GM commercial, and from a Lexus commercial, and a Nissan commercial, and … Not only do they all look alike, they’re all goofy come-ons with music the $16 million/year agency couldn’t even be bothered to write themselves. Most advertising now seems designed to trick the customer into engaging with it, rather than to deliver a relevant, well-timed, and entertaining message.
For an excellent and very current example of this trickery, one need look no further than the industry’s current fervor around ad blocking. As the Direct Marketing Association‘s Neal O’Keefe points out, the response to customers saying “I don’t want to see your message” is to figure out how to make them see it anyway, with tricks such as “re-insertion”.
The feel they need to do this because subconsciously they realize that viewers of online content have developed a stigma of online advertising: that it’s going to suck, and that it’s going to be interruptive or intrusive.
So they treat the symptom. Rather than trying to make the advertising content, well, good, they spend millions of dollars trying to come up with technologies that will make you see it anyway. It has become an arms race; each side alternately outgunning the other with technology tricks.
The solution seems simple enough: make people want to – rather than have to – engage with the ad. Of course, this isn’t always simple, but that’s a different post.
2. Downward Price Pressure on content
It’s a familiar story by now: the Internet has forced publishers to rethink their “circulation revenue” model. It was never really easy to optimize “Who pays for content, and how much”? The Internet added a whole new dynamic, and the overall effect on content end user price is negative – at least for the time being. This has certainly come with some well-publicized tradeoffs, but that, too, is another post.
3. Customer-Centric Blinders
Somewhere around the mid-1990s, “customer-centric” became one of the prevailing management fads. This is not to say that it was as vapid as many management fads end up looking in retrospect; quite to the contrary. As we entered the data-driven world, the idea of understanding a company’s ecosystem – with the customer placed at the center of a network graph – became possible in ways it had never been before. (Or at least not since those networks consisted of just the other 20 members of your tribe, and maybe the three neighboring tribes.)
It also established a heuristic that’s still extant today: the customer is the center of the universe. This is certainly not a bad thing. Placing the customer at the center of processes and analyses facilitates optimizing for the customer, the source of revenue, which is arguably the most important type of cash flow in a going concern.
Take, for example, this visual representation of the Northwind database, showing one customer (Island Trading), the entire network of products they bought, and the original suppliers of those products. Each arrow in this graph represents some relationship or process:
To be clear: placing the customer at the center of processes is a good thing for many reasons. My argument, though, is that in an age of data & location, we need more than just a customer-centric view of business. Locations are data-rich, and people interact with locations in complex ways.
4. “Context” is Missing from Location Data
Privacy concerns may be responsible for some of the foot-dragging in location data. It’s no secret that both publishers and advertisers are collecting vast amounts of data on who were are, what we read, who we know, etc. But they do have to be somewhat careful about how they use this data, lest they be convicted in the court of public opinion – or even an actual court. My skepticism about Facebook’s concern for its users’ privacy notwithstanding, a friend who is an engineer there assures me that the company has evolved in its view of users and that the Federal Trade Commission “sits in their lap” anyway. (I’m still skeptical.) European regulators are even fiercer protectors of consumer privacy than the FTC.
Publishers and advertisers have not really advanced beyond “Who is at that specific point now?” and “What is at that specific point now?”
What’s missing is “Why are they at that specific point right now?” Or, not really missing, per se, but there’s a default assumption about “why?” that’s treating it as a very blunt instrument, because it is centered around a specific point at a specific time:
- “You are shopping at Macy’s now. Would you like a Macy’s coupon?”
- “There’s a Starbucks near you! Come on in for a free sample of our new vanilla spiced pumpkin chai soy latte, now with only 12 teaspoons of sugar!!”
I used to joke that I could be walking by a custom bra boutique on the Upper East Side of Manhattan, and if they were running a sale, I would get an alert about it, despite being a rather obvious non-bra-customer.
In publishing for location-based media, context has effectively been supplanted by location; we now match users to where they are rather than what content interests them:
And yes, this could be down to the inability to use consumer data effectively with location. But remember, in the “old days”, magazine publishers didn’t need any consumer data to put the right full-page ad opposite the right article. They understood the consumer through the context of the content. What this should look like is:
But this still won’t solve the problem the way location data is currently done because we’re still thinking in terms of point content! So we’re left with a status quo where location data is under-serving publishers, advertisers, and consumers. With a better use of location data, we could make content that engages consumers better – whether that content is advertising or “editorial”.
So somehow we want to derive “why is this person at Macy’s?” or “whyis this person passing that Starbucks?” from a context that has to do with the location.
Suppose my friend Shara comes to New York City to visit. We’ll get together during one or two of the days she’s here, but I have to work too. So she’ll be on her own some days. That’s fine, because a) Shara’s into solo urban exploration, and b) her Airbnb host, Jessica from the East Village, left a handwritten note that included a list titled “A Perfect Tourist Day in New York”. While I don’t think it is anywhere near as superlative as its title claims, it’s an OK way to kill a few hours on a perfect day on the Lower East Side. I re-created her list here:
Already, there’s a lot of information there. For example, we know:
- Jessica recommends each of these places
- Each of these places is recommended by Jessica (yes, there’s a difference! I recommend looking into Graph databases.)
- Jessica thinks of these places as 7 members of a cohesive group
- The site of the John Varvatos store used to be CBGB (an iconic – and historic – music club of the ’80s)
But wait. We already know that the John Varvatos store is at the former location of CBGB:
I also happen to know that CBGB, though it is no longer at 315 Bowery (and somewhat tragically, is now a restaurant at the Newark International Airport, and you can get Disco Fries there), is a stop on Kevin Stein’s fantastic Rock & Roll History Tour of Greenwich Village. So someone who’s visiting 315 Bowery might be a fashion slave, or a punk rock fan, or a history buff, or an Airbnb guest of Jessica’s. All of this based solely on what I know about the location, which in turn is based on that location being in just two different lists. (Imagine if we really start layering in things we can know about that location, if we’re not constrained by what is there now, and only on the surface of the earth.)
This is the idea of location-centric. It does not need to replace customer-centric frameworks; it can augment them. Modern Graph Databases are perfectly suited to this use. (I used Neo4j, an open source Graph Database, to generate the network diagrams in this section.)
Context Through Location
By adding location-centric modeling and heuristics, publishers across a broad spectrum of end-use applications can derive context without heavy reliance on PII – Personally Identifiable Information. This is perfectly analogous to the original context-driven advertising in early magazines and newspapers, and it works to this day.
Moving to this model, though, will require purpose-built databases that connect location to location based on some shared context. That shared context, though, may be something other than “what is at this point now?”
A heavy lift, to be sure. This is, however, not as difficult as it may sound. Much of the information already exists in innumerable and highly detailed GIS databases, and with modern “big data” tools, can be pulled together. (With a bit of work, yes. But it is actually achievable.) And location data can be crowdsourced as well. Disclosure: this was a major plank in the Moveable Feast Mobile Media strategy.
One final note – on Pokémon Go. As I mentioned at the top of this post, the craze over this new Augmented Reality((some “experts” would argue that Pokémon Go is not true AR. While I do not agree, I think I’d be remiss in not pointing that out.)) game indicates a hunger for location apps that include a context. Here, the game – essentially a scavenger hunt – is the context. The point, though, is that it brings end users to places because of what they are doing. It connects those places to each other. And the implications are huge, even if the game itself turns out to be a passing fancy, which I think it will.