Instead of kicking data brokers off its platform, Facebook should empower its entire user base to be their own brokers of data.
Late last week Facebook announced it would eliminate all third-party data brokers from its platform. It framed this announcement as a response to the slow motion train wreck that is the Cambridge Analytica story. Just as it painted Cambridge as a “bad actor” for compromising its users’ data, Facebook has now vilified hundreds of companies who have provided it fuel for its core business model, a model that remains at the center of its current travails.
Why? Well, I hate to be cynical, but here’s my answer: Because Cambridge Analytica provided Facebook air cover to consolidate power over the open web. Put another way: Facebook is planning to profit from a scandal of their own making.
But more on that in a moment. For now, I want to dig into this “third party data” issue, because in the end, Facebook’s announcement amounts to an insidious head fake that deserves to be called out. It also cements Facebook’s commitment to a bankrupt philosophy that runs counter to the fundamental goodness that is the Internet.
Yes, I’m angry.
I’m so angry, in fact, that I waited three full days before even attempting to write about this issue. I’m also utterly conflicted, at least financially, because I am on the board of two companies that are either directly or indirectly impacted by Facebook’s decision. But if you think I’m expressing this opinion because I’m trying to make (or protect) my financial interests, please move along. I am engaged with two companies that reflect my philosophies and my motives, both of which have been formed well before I joined either company’s board. Facebook has never been a friend to either Sovrn, which runs an independent publisher service platform driven by open data, or Acxiom, which has built the most important platform in the world for the interconnection of data across disparate sources. Both are enablers of what I’ve long called “the independent web,” a concept that it seems is ripe for a renaissance.
There’s a lot to this post, and I even considered writing it as a multi-part series. But instead, here’s a short guide to what is a relatively long post (for today’s Interwebs, anyway).
- A primer on data brokers and their relationship to “third” and “first” party data.
- Facebook’s role as the biggest data broker of them all
- The core issue of trust, and
- A straw man for a better model for Facebook and the web.
So, pour yourself a bourbon folks, and here we go…
Understanding Third Party Data
Late Weds evening Facebook posted one of the shortest announcements I’ve ever seen. Here it is in its entirety:
Shutting Down Partner Categories
We want to let advertisers know that we will be shutting down Partner Categories. This product enables third-party data providers to offer their targeting directly on Facebook. While this is common industry practice, we believe this step, winding down over the next six months, will help improve people’s privacy on Facebook.
Comments were turned off for this post, a decision I am certain was intentional. Given the context of the Cambridge crisis, the implications are clear: Third-party data providers are bad actors, and cutting them off will “improve privacy on Facebook.” “Partner Categories” is a euphemism for “targeting data that’s not from Facebook.”
So what is this “third party data,” and why am I so worked up about all of this? In short, it’s consumer information collected and packaged by a vast ecosystem of companies. This information powers the advertising-driven model of the modern Internet. As I’ve written over and over, that model is messy, often abused, and deeply in need of reformation. However, it is also the only model we’ve got, and tens of billions of dollars flow through it, driving revenue to hundreds of thousands of independent publishers who depend on it.*
Companies who specialize in packaging third party data are often referred to as “data brokers,” and they are regularly vilified as evil practitioners of “surveillance capitalism.” Acxiom is one of the best known and most misunderstood members of this group. The company purchases rights to public or legally sourced information — census, mortgage, and other data — and creates privacy secure, anonymous “segments” that marketers can use to target customers across various digital channels, including Facebook (until now). Acxiom has many other businesses as well, including LiveRamp, which allows marketers to “on board” their own “first party” data and use it across the ecosystem. Acxiom’s future is clearly aligned with the LiveRamp business, but I’ll get to that in a minute.
I’ve been on the board of Acxiom for nearly five years, and to be honest, I was stunned the company even wanted me there. I’ve a long history of railing against the current state of the data ecosystem, and I don’t ask permission before I post my musings. I led the initiative against fraud in the adtech ecosystem, and the resulting policies eventually became mandatory across the industry. But in my thirty years of engagement with the marketing and media industries, I’ve never met a more serious, ethical, and well-intentioned group of professionals as the people behind Acxiom. They’re obsessed with the security, privacy, and legal compliance of their data business. They have to be.
Earlier I used the phrase “first party data,” and it’s important that we understand what that means. Third party data has a major image problem: It’s collected without the direct permission of the people who create that information. As I said above, some of it is already public information, and therefore fair game. Other data is collected by first parties — companies that have a direct relationship with their customers — but in a fashion that allows those first parties to sell that data to third parties. This is the case with financial information, for example. Equifax’s IXI Service purchases aggregate income and investment information from financial services companies who have a first party relationship with their consumers. Equifax then repackages that information as income segments which help marketers target consumers based on their inferred household income. The first party — your bank — makes money off your data, and the third party — Equifax, in this case — offers the data service and takes the rap if Facebook decides to vilify an industry.
So why did Facebook throw Acxiom, Equifax, Oracle, and others under the bus?
Facebook Is The Biggest Data Broker In The World
As the largest data collection platform in the world, Facebook has more first party relationships than any other entity in human history. When you agree to use the service, you agree to the company’s terms of service, which cement a first party relationship, and give Facebook the right to use your data to drive its advertising business, among other things.
But until recently, Facebook lacked some key data points. You might tell Facebook your relationship status, your friend network, and the cat videos you like, but you don’t reliably give Facebook information about your income, the cars you purchase, your ethnicity, your political preferences, or your home address. That was a problem for the company as it sought to build its core advertising business — advertisers like to target people by these characteristics. So Facebook sought out partnerships with the best of breed companies who had that information, and executed deep integrations between those companies and its own advertising platform. That allowed marketers to target their customers on Facebook just like they target them across the open web (or on television, for that matter).
But over the past few years Facebook has gotten better and better at inferring these key data points about their own customers. The longer we use Facebook, the more data we give its modeling algorithms. We may not declare we’re Republican on our profile pages, but if we “like” enough conservative pages, Facebook is going to figure that out. We may not declare our household income on our profile page, but if Facebook notices we keep leaving the site to visit FirstRepublic.com or Walmart.com, it’s not a stretch to place us in an appropriate income bracket. Over the past five years, Facebook has perfected the art of inferring just about everything about you. It has a vast network of its own agents of “surveillance capitalism” strewn all over the web (all those Like and Share buttons serve multiple purposes), and it employs some of the best minds — and the best machine learning — in service of its core business model of collecting data about you, then selling access to that data to advertisers.
Not to mention, Facebook has certainly been watching how all those third parties are working across its networks, and it alone has access to all that information, which it alone can use to model replacement products if and when it deems it’s profitable to do so.
All of which means exactly one thing given last week’s news: Facebook has just become the biggest data broker in the history of humanity. It just doesn’t want you to know that.
So What Does This Have To Do With “Privacy”?
Let’s return to Facebook’s announcement about kicking data brokers off its platform. The key reason given was in that last line: to “help improve people’s privacy on Facebook.”
Taken literally, I suppose that means that all those third-party data segments that were previously used to target advertising on your feed are infringing your privacy. This of course means that Facebook will no longer target you based on that kind of data — data like your income, your political affiliation, your propensity to be in the market for an automobile, or your zip code.
Ummmm…no. This just means Facebook has gotten good enough at inferring that same data, and it no longer needs third party partners to help it satisfy advertiser demand. It’s watched the behavior of all that third party data on its network, and it’s gotten good enough to rebuild exactly the same thing inside its own walls. It can now own the entire profit chain.
Your “privacy” has pretty much nothing to do with this decision.
This story from the Washington Post gets to the core of the issue — by kicking third party data providers off of its platform, and conflating them with the Cambridge mess and casting them as “bad actors” in the process, Facebook is creating a homogenous, anodyne world driven entirely by…Facebook.
This is a brittle and dangerous approach to the public square. In the name of “privacy,” Facebook has declared that data from sources other than Facebook itself are suspect and not worthy of transacting within its walled garden. This means one thing: Facebook sees itself as a totality, a universe unto itself, unwilling to engage in the kind of free and open exchange that has characterized most of Internet (and human) history.
And that, my friends, spells the beginning of the end for Facebook.
Who Do You Trust?
Facebook has always been a walled garden, despite Zuckerberg’s brief flirtation with all things open web back in the early days. I’ll never forget a dinner I had with a senior Facebook executive five or so years ago, about the time I started warning anyone who’d listen about the dangers of placing your faith in a platform that essentially controlled the weather. We debated the relative value of the open web versus Facebook’s closed platform. As I brought up example after example of things that work well on the open web — publishing your views, starting a new online business, debating politics in a forum, etc — this executive simply countered each of them with one response: “You can do it on Facebook, but better. Why do it anywhere else?”
I’ll tell you why. Because it’s impossible for one company to fabricate reality for billions of individuals independent of the interconnected experiences and relationships that exist outside of that fabricated reality. It’s an utterly brittle product model, and it’s doomed to fail. Banning third party agents from engaging with Facebook’s platform insures that the only information that will inform Facebook will be derived from and/or controlled by Facebook itself. That kind of ecosystem will ultimately collapse on itself. No single entity can manage such complexity. It presumes a God complex. If I were a Facebook shareholder, I’d be very worried right about now.
If you take Facebook’s expulsion of third party data brokers to its logical extreme, then Facebook should ban all “foreign” data from entering its platform. But of course, the company isn’t doing that at all. It’s happy to welcome that aforementioned “first party” data onto its platform for targeting purposes (IE to drive its core business model of advertising). Retailers like Target, for example, have tons of first-party-derived data about their customers that they gather from purchasing behavior, loyalty programs, and visits to their website and app. Target keeps this data in a massive database, and it rightly considers this data to be proprietary and fundamentally central to its business.
Under Facebook’s new regime, however, Target must now upload that data directly to Facebook, without the help of a third party like Acxiom, which provides a “data safe haven” for first parties to upload their data to third party sites like Facebook while retaining control over how that data is used. (This is where Acxiom’s LiveRamp business comes in). But under Facebook’s new rules, Acxiom is kicked to the curb, and Facebook alone will have direct access and control over Target’s data. By extension, that means that Target must trust Facebook to never misuse Target’s most sensitive and important data to, say, help train Facebook’s advertising algorithms to benefit other retailers who compete with Target, or to, I don’t know, game the entire retail sector to the benefit of Facebook shareholders. And we all know we can trust Facebook to never use data to its own advantage, right?
Not. Gonna. Happen.
Are you starting to see what I mean by “brittle”?
A Better Model For The Web, And For Facebook
So if Facebook’s core model is brittle and doomed to fail, what might be a better approach? Here’s the core of my response: Embrace the true meaning of “platform.”
Now hold on, you might respond. It was Facebook’s flirtation with being a platform that created the entire Cambridge Analytica problem in the first place, no? Allowing third parties to create applications on Facebook’s platform created the Frankenstein we’re now struggling to contain!
Well, yes, but let’s not toss the baby out with the bath, shall we? A true platform is neutral, allowing all manner of actors to engage across it to connect and share value under a clear and consistent governance framework. The failure of Facebook’s platform was not a failure of the concept of a platform. It was a failure of Facebook’s governance of that platform.
If Facebook truly wants to address its most fundamental flaws, it should adopt an approach which is exactly the opposite of its current course. To become a neutral platform, Facebook should not build the garden walls higher. Instead, it should re-commit itself to being an open and neutral platform for the exchange of value across not only its own services, but every service in the world. (Bill Raduchel has promoted a version of this, covered in this Information article on Friday).
If it wants to truly be a neutral platform, and this is crucial, Facebook has to get out of the business of selling media. This, of course, represents nearly all of the company’s $40 billion in revenue, but if the company wants to be a data broker, which it already is, it can’t also be in the business of selling media. That creates insoluble conflicts of interest. Instead, it should open its network up to all comers, make its data available to everyone (including its users), let everyone compete to sell media on its platform (and using its data, across the web), and take a small percentage of all that revenue in a neutral and non-biased fashion. It’s entirely possible that this model would ultimately yield far more revenue than Facebook’s current top line, and prove much less brittle to boot.
In short, Facebook should tear down those walls and open itself up to all comers, but only under a robust system of governance informed by multiple stakeholders — consumers, governments, businesses, academics, and more. To do this, it would have to dedicate itself to several core principles, all of which are, unfortunately, anathema to the company at present.
- The user in control. Put each and every Facebook user under the governance framework of a Data Bill of Rights. Facebook could co-own all data created by that user, but also allow that user to to create value off platform with that data. And sure, Facebook can make money with this model — perhaps a vig which goes back to Facebook for every transaction performed off platform.
- Data portability. Related, Facebook should allow every single Facebook user to export a machine readable version of their data to any service they chose, easily and without friction. That’s what “co-ownership” means. Yes, this means competitors will flourish. But that will only force Facebook to be the best at creating value with the data it has helped to create (and partnering with those potential competitors in new and unique ways, again which could bring in revenue).
- Identity management. Facebook is already one of the largest and most significant identity management services in the world. But it exists to serve just one entity: Facebook. There is a huge opportunity for the company to become the defacto platform for identity worldwide, but only if it manages that sensitive identity data in a truly neutral fashion — which is, alas, anathema to how the company currently operates.
- Third party auditing. Submit all data and processes in the company to an independent, trustworthy third party that ensures Facebook’s new governance model is being held to account. Has the world ever seen such an entity? No. Do we need one? Hell yes.
I’ve hinted at this model in previous posts, and in a future piece I promise to expand on it. But that should do for now. The idea is simple: Tear down the walls that separate Facebook from the rest of the world, and instead build a platform that enables all of us to be in control of our own data, enabling in the process not only untold value for Facebook, but for the world at large. This requires we pivot from a model where all of our data is controlled by one or two or three large platforms (oh hai, Google and Amazon), and is instead controlled at the node — the individual (and the myriad businesses that would spring up to leverage that data). This is a radical shift, but in the end, it’s the only model that will work at scale.
Of course, there are other approaches. Facebook could continue with its core business of exclusively selling its own media, but if it does, it should abandon its role as a data broker, and let others do that work by freeing its users data to be brokered between the users and qualified third parties (like the ones it just threw out). This would free it from the conflict inherent in practicing the dark arts of advertising targeting, the very same arts which are currently laying blame for democracy’s demise at the company’s feet. This scenario deserves its own post, but again, such a move is exactly the opposite of what was just announced.
In either case, it’s simply unsustainable to expect that one company can manage the complexities of our data-driven future. It takes an ecosystem to build a civilization. The sooner Facebook realizes this, the better the outcome for humanity.
Your move, Mr. Zuckerberg.
* I have a deep history with the birth and adolescence of internet advertising and adtech. I’d be happy to tell you about it sometime.