What Facebook Knows and Doesn’t Know

Privacy concerns have been in the forefront of the news lately, and no article discussing privacy is complete without mentioning Facebook. That’s because Facebook is considered to be the all-knowing machine that’s tirelessly collecting data about us and turning it into insights that can be used to better market things to us with extreme precision. Certainly Facebook isn’t the only online juggernaut with this strategy and sophisticated data collection capabilities, but in many ways it’s the poster child for our collective concerns and anxieties.

I joined Facebook in 2007. At the time, it was becoming the next big thing, and I wanted to see what it was all about. After some initial excitement, I noticed my usage dropping as the years went by. My usage massively dropped in 2019 when I somehow changed my default language settings to German and I didn’t feel any real urgency to figure out how to undo it, all this to say I am certainly not a typical Facebook user.

While not a high intensity Facebook user, I am a high intensity data nerd, so when I read an article that explained how to peek under the hood to see in detail what Facebook knows about you, and what it has learned about me from third parties, I of course could not resist. If your interest is equally high, start your journey here: https://www.facebook.com/off_facebook_activity/

I clicked all the options so that I could see everything Facebook knew about me. While not a heavy user, I was a long-term user, and I imagined Facebook had likely learned a lot about me in 13 years. In due course, Facebook presented me with a downloadable Zip file that contained a number of folders.

The folder “Ads and Businesses” turned out to be the money folder. This is where I learned my personal interests as divined by Facebook – all individual categories that can be selected by marketers. Here are some highlights of my interests:

• Cast iron (who doesn’t love cast iron?)

• Scratching (what can I say?)

• Tesla (Facebook helpfully clarified that my interest was not in the car, but rather the band … the band?)

• Oysters (I don’t eat them)

• Skiing (I don’t ski)

• Star Trek (absolutely true – when I was about 14 years old)

 There were about 50 interest categories in all; not all wrong, but overall far from an accurate picture. What I infer by looking at these interest categories is that they are keywords crudely extracted from various ads I had clicked on over the years. I say “crudely” because these interest tags don’t represent an organized taxonomy; there is no hierarchy, and there is only a lackluster attempt to disambiguate. For example, one of my interests is “online.” Without any context, this is useless information. And if Facebook assesses the recency of my interests, or the intensity of my interest (how many times, for example, did I look at things relating to cast iron?), it is not sharing these data with its users.

If Facebook underwhelmed me with its insights into my interests, the listing of “Advertisers who uploaded a contact list with my information” totally confused me. I was presented with a list of literally hundreds of businesses that ostensibly had my contact information and had tried to match it to my Facebook data. What I saw on this list were probably close to a hundred local car dealerships from all over the country, followed by almost as many local real estate agencies. I feel certain, for example, that I have never visited the website of, much less interacted with, International Honda of Sheboygan, WI. But this car dealership – reportedly – has my contact information and is matching it to Facebook.

There are a few possible explanations for this. The one I find most likely is that in the case of automobiles, some unscrupulous middlemen are selling the same file of “leads” to unsuspecting car dealers nationwide. It could also be inexperienced or bad marketers or marketing agencies. Some free advice to Toledo Edison, Maybelline, The Property Girls of Michigan, Bank Midwest and Choctaw Casinos and Resorts – take a look at your list sources and maybe even your marketing strategies, because something seems broken.

Looking at your own Facebook data gives you a rare opportunity to see and evaluate what’s going on behind the curtain. To me, Facebook’s secret sauce really doesn’t appear to be its technology. Grabbing keywords from ads I have clicked is utterly banal. Offering marketers hundreds of thousands of interest tags does in fact allows for extreme microtargeting, but in the sloppiest, laziest possible way. Capturing all my ad clicks is useful and valuable, but hardly cutting edge. What appears to make Facebook so valuable seems not to be the data it has collected, but the fact it has collected data on a hitherto unknown scale. Knowing that I have an interest in flax (yes, this is really one of my reputed interests!) even if true is pretty useless until you get enough scale to identify thousands of people interested in flax, at which point this obscure data point suddenly acquires monetary value.

What my Facebook  data suggest is that while it may not be good enough to deliver the precision and accuracy many marketers have bought into, what it has done is create “good enough” data at extreme scale. And that is proving to be even better than good enough. 

Thinking About Privacy and Data? Good.

We have heard a lot in the past few weeks about the travails of Facebook, as it became widely known that many millions of its user profiles had been,  for lack of a better term, hacked. That in turn brought Facebook’s advertising microtargeting capabilities into focus, creating more widespread privacy concerns.

But does the average data publisher have to worry about privacy? The short answer is yes.

Data publishers, including B2B data publishers, often control a wealth of extremely valuable data. Many data publishers don’t fully appreciate what valuable insights they could glean from their own data. Fortunately, data thieves haven’t figured it out either … yet.

The highest value data in a typical commercial database isn’t the data itself, it’s the information on what users are doing with the data. Knowing, for example, that the head of acquisitions at a public company was doing deep research on another public company, could be extremely valuable to certain people. Knowing that an executive suddenly started looking at job openings could be valuable. Knowing that five venture capital firms in three days had looked up information on a particular start-up could be extremely valuable. You get the idea.

We already sell some types of information about how users interact with data, and we do this with very little thought about how it might blow up in our faces. Other of our data is clearly quite sensitive and we’d never sell it, but what if somebody stole it?

Going back to 2013, Bloomberg came in for tough public scrutiny after it was revealed its reporters had used Bloomberg terminal access data to track an individual in order to write a story. That’s pretty tame compared to the recent Facebook revelations, but it shows there is often tremendous inferential data hiding in the intersection between our databases and how our customers interact with it. Monetize where appropriate. Protect where appropriate. But whatever you do, don't ignore it. 

LinkedIn's New Corporate Directory

In my view, the future of LinkedIn depends on finding ways to get itself inside of business workflow – the essence of infocommerce – because the history of databases that remained standalone reference products is a sad one.

LinkedIn’s first big push to build ongoing user engagement was to add user-generated content, lots of it, creating a B2B Facebook if you will. This is certainly a valid approach, but with the Internet already groaning under the weight of endless content, much of it free, this is a tough road. I think workflow integration is a lot easier and ultimately much stickier. It is, fundamentally, the difference between logging into LinkedIn to “stay current” or perhaps find a useful morsel of information through sheer serendipity, and logging into LinkedIn because you need it to do your job.

Well, LinkedIn took a small but important move in the direction of workflow this week with the launch of LinkedIn Lookup. Very simply, this new app allows you to turn LinkedIn into an internal company directory.

As you can imagine, if you were to filter all LinkedIn profiles by current employer, you would essentially get an internal company directory. And it would be better than almost any company directory that exists given the depth of its profiles and the high level of data accuracy. But the new LinkedIn app does more than just filter listings, it also prioritizes fellow employee listings over your own connections so you’re really using it as an internal directory. Corporate email addresses are shown as well.

Overall, LinkedIn Lookup is a fairly weak version 1.0 app. But if LinkedIn sticks with it, it could take this product in some very interesting directions. Consider:

·        Setting up the product with a corporate administrator could help make listings more accurate (many people don’t update their employer information if they are not immediately going to a new job). In addition, LinkedIn could make this administrator the point person to maintain the company web page as well, helping to insure deeper and more accurate data

·        With listings now used for employment purposes, employees will be more diligent in maintaining their listings to the benefit of both the company and LinkedIn

·        By letting employees see all the connections of other employees, an extremely powerful networking tool along the lines of those offered by Reachable can be offered.

·        Non-public fields could be made available for corporate directory purposes such as reporting relationships, and this could in turn enable real-time organizational charts

·        The product could offer links to a company’s payroll system (as many internal company directories already do), to help insure even higher levels of accuracy

And that’s just a starting list. Indeed, an enormously powerful product platform exists for LinkedIn to exploit with only some additional programming effort. And this product, properly evolved, is certainly one LinkedIn could charge for. No company wants to maintain its own internal directory if it can avoid doing so, and LinkedIn would bring to the table features and functionality no company could duplicate on its own because of its connections data.

Best of all, as companies adopt LinkedIn as their internal directory platform, LinkedIn automatically becomes a stronger database as a result. Employees who haven’t yet built a profile will do so; and those with existing profiles will be motivated if not required to keep them current.

Sure, there are some data governance issues that will need to be addressed and doubtless some technological and structural bumps in the road will emerge; as the saying goes, “hierarchies are hell.” But these issues will come to the fore because LinkedIn is simultaneously becoming more important and the end result of that is a more comprehensive and accurate database for LinkedIn, that will give it the basis to chase even more data-driven workflow opportunities.

If LinkedIn wants to offer high quality user-maintained data that gets accessed frequently, there’s no better way than to help it enable daily business activities. LinkedIn Lookup can be an important first start in this direction.

The Value of Volume

A fascinating story in the New York Times takes us inside a planning session between Facebook and a major consumer products marketer, Reckitt Benckiser (RB). RB wants to market its fish oil nutritional supplement via Facebook. After listening to Facebook executives pitch a large and broad marketing campaign, the RB marketing manager stops the conversation to note that he’s interested in using Facebook because of its targeting capabilities, not because of its massive reach. He suggests that perhaps RB would be better served by targeting known fish oil buyers as well as buyers of other products that are suggestive of the consumer having an interest in heart health – a key benefit of taking fish oil.

The representatives from Facebook, rather than embracing a targeted approach, instead pushed back. It wasn’t that Facebook couldn’t deliver this highly targeted audience – it could. The primary objection of the Facebook team was that a highly targeted marketing campaign like this would be “too expensive.” And keep in mind that RB is a major global marketer, with annual sales pushing $15 billion.

If you find it odd that Facebook is pushing marketers to broadly-based marketing campaigns, the underlying logic is even more intriguing. The Facebook “advertising strategist” in the meeting explains the company’s view by saying that advertising on Facebook was like firing a shotgun. My early education in the world of direct marketing taught me that the goal was to use information to move away from sloppy, imprecise shotgun marketing, to precise rifle-like marketing. Perhaps this strategist simply butchered this well-known metaphor. But as you read deeper into this article, you start to see that Facebook really wants to sell big, volume campaigns with only minimal targeting.

You probably see where I am going with this. Facebook develops this truly massive and engaged audience. On top of this, it has unbelievably detailed and accurate information about this audience, and can target against these data. Yet when a large and sophisticated consumer marketer wants to take full advantage of this deep targeting capability, Facebook proposes a mass media solution instead. Sure, Facebook is happy to tweak this huge campaign once it rolled out, but Facebook clearly wants you to start big and whittle your campaign down over time.

Why is Facebook seemingly disavowing the precision targeting that is ostensibly its greatest asset and differentiator? While the article doesn’t provide any numbers, the likely answer is this: money. You make more money selling volume than precision.

I saw this many years ago in the direct marketing world. Everyone there preached the gospel of targeting – rifle precision over shotgun sloppiness. I listened to endless presentations by clever people showing how you could use data to identify and reach the exact best prospects for your product. But try to actually purchase these highly targeted names, and you got the exact same tap dance now being performed by Facebook. The reality is that nobody with a ten million name database wanted to sell you just 500 of those names – your absolutely best prospects. Direct marketing was a volume-based business: you made your money selling big lists, not precisely targeted lists (which truth be told were kind of a hassle to produce anyway). It was a business where you talked targeting and sold saturation.

And lest you think I am pointing a finger only at media companies, be assured that marketers need to sign up for their fair share of the blame. For example, I see B2B media companies working so hard to deliver fresh, hot, highly qualified sales leads to marketers who in many cases won’ t buy them unless the media company can guarantee a minimum volume each month. Quantity trumps quality – again.

Like it or not, there’s an important lesson here for those of who run media and data companies. Precision and targeting are great – but only to a point. Marketing and sales activities are inherently sloppy and imprecise activities, and that’s why they depend on volume. And we also now must acknowledge that in a digital world, mass marketing is cheap and can often yield powerful market research insights. That’s why, despite all the advances that have been made in increasing precision, there will be continuing value in volume.