The Blind Spots in Facebook's Ad Machine

was seen as embarrassing for senators. How could they be so clueless? A lot of people know Facebook's business model is advertising, but they may not know exactly how it works. In for example, when Facebook disclosed that Russia purchased a hundred thousand dollars and ads on Facebook to manipulate US politics, I got this question from a lot of smart people. Who was the Facebook salesperson that sold Russia those ads?

Can we find them? That sounds like a pretty reasonable question, But when Facebook employees here that story, they immediately laugh because Facebook doesn't have sales people directly selling ads. In most cases, it all happens through a website. Anyone with a credit card can do it, and you don't even have to be in contact with Facebook at all, just the Facebook website. You don't even have to be who

you say you are. So much has happened since that revelation, so many changes to Facebook's advertising system, so many apologies, so many more questions from Congress, And even that didn't stop Facebook from having its share of crises in the last election cycle as I tried to reform the system.

But if there's anything this era of Facebook News has proven, it's that sometimes it's worth it to try to understand the basis by I'm Brad Stone and I'm Sarah Fryar, and this week on Decrypted, we're going to try to explain how Facebook's advertising system works so we can understand why some people love it so much and others think it could be so dangerous. Facebook's ad machine is easy and quick to use, and that's made it incredibly profitable. Wall Street is expecting it to bring in fifty five

billion in revenue this year. But the same algorithms that make it so powerful can also create problems. For example, if enough people say on their profile that they hate Jews, Facebook software would make a category all by itself to let advertisers send their contents specifically to do hitters, and unfortunately that's not even a hypothetical. That's one of the problems Facebook had to solve, along with a lot of

other things. But fixing oversight for the ad system isn't as easy as you might think, and some of those updates have had unintended consequences of their own stay with us. So Sarah back to that congressional hearing in April, the senators asked Mark Zuckerberg about whether Facebook sales user data to advertisers. Um, have you ever drawn the line on selling data to an advertiser? Yes, Senator, we don't sell data at all. So the way the add system works

is advertisers can come to us. The company hates it when people think that because it doesn't sell data to advertisers. What it sells is access to certain kinds of people based on the information they share on face and then we help them get that message in front of people. But this is one of the it's widely mischaracterized about our system that we sell data, and it's actually everything you share on Facebook, all the articles you decide to click on or decide not to click on, everything you

write in that textbox. All the things you're friends like help paint a picture of who you are. Facebook knows what stories you go to, and it knows as you're rolling what you stop to stare at, even if you never click on it. All of these things end up as signals that Facebook can you use to determine what kind of ads a person might be interested in seeing. Are they rich or poor, Asian or Latino, a mother

or a student. The advertisers can't really see who is getting their ads, and Facebook can't at an individual level at least either. All they know is it works. The fact that Facebook doesn't really sell it's data, it just sells access to it is actually one of the reasons it's been so successful. It has all this information on now more than two billion people basically locked up, and it's only available to people or groups advertising directly on Facebook.

And it's actually one of the most important parts of how Facebook works is we do not sell data. Advertisers do not get access to people's individual data. Those advertisers can use Facebook system to get really specific about what kind of people they want to reach. If enough people are into something, Facebook will auto create category for it,

like breeding insects. It might be kind of hard to reach bug breeders if you're advertising in the New York Times, But on Facebook you can let exactly the right people know about your upcoming beekeeping conference. But what if you want only white people to apply for your apartment complex, or only young people to apply for your open job. Facebook, for a time actually allowed people to do that too. Of course, that's not how Facebook intended for the ad

machine to work. And while there are some awkward, unintended consequences on the whole, the ad platform works very well for a lot of its users. Like a couple of months ago, I went with our producer Piaga Kari to visit a small business in New York. It's called Two Blind Brothers and they make T shirts. Should we give you guys quick toward? Yeah, let's do it. Um, well, you're pretty much looking at all of it already, but

we have a few fun things, and here we have this. Uh. The company is run by two brothers, Brad and Brian. They're both visually impaired, and profits from their T shirt sales are donated to organizations that fight blindness, working on some stuff for kids. Uh, we have t shirts Brad and Brian say their company would never have even gotten off the ground without Facebook's ad system, which allowed them to start boosting posts on their business page without spending

a lot of money. It works so well that Facebook actually published a case study about one of their Black Friday campaigns. The actual study is there on a wall in a frame, and it's been signed by Chief Operating Officer Cheryl Sandberg herself. So why did you start advertising on Facebook? You know, Facebook does a great job of baby stepping you into experimenting with the adds. So I remember we had put out this video and we went to bed that night. We woke up the next day

and we're like, what the like? Normally when I put out a post on Facebook, it's like Grandma and like my sister that like it? And then this post just and So they had this dynamite video and they started targeting certain audiences with it. They could select for specific age groups or find people with an interest in blindness. But the brothers said it actually worked better when they could just click a little button that would get Facebook to do the job for them, choosing on its own

who could see their ads right. They started using a tool called look Alike audiences. Yeah, so you know I can. I can get on the app platform and I can literally put into the to our audiences. I'm looking for somebody who uh likes tom shoes and like selling Degenerous because you know, we're featured on her show, and who is an online shopper and all these sort of broad

categorical um characteristics. Or I can let Facebook analyze who's interacting with our brand and then they can decide on hundreds of factors using all their technology to figure out who that best audience is. But that is an audience that Facebook has determined for us looks and acts like our best and most exciting customers, and then we can take our ad creative and send it to that audience. So the lookalike feature just finds people who are similar

to the users you've reached already. It's all automatic, and as the Brothers realize pretty quickly, it's incredibly effective. It's not just effective for businesses, though, it's effective for all kinds of advertising. This type of ready made targeting was one of the main tools Donald Trump used in its presidential election campaign. It allows someone who might not be in the mainstream conversation to find just enough people who share an interest, and then people who look like those

people and expanded from there. So we know that Facebook's ad targeting is super efficient, and we know that businesses and political groups are using it. But a lot of these tools still feel somewhat mysterious to a lot of people. So we went to Facebook's headquarters in Menlo Park and asked them, Yeah, Thursday traffics bad. Bad. It was like telling me that I was going to get here. Hi, hi Hi. This is Nikila Shwinavasan, the product manager for

Facebook Small Business Ads Growth. She knows inside and out the portal that most people come into when they want to buy ads, and she also knows that not everyone understands how it works behind the scenes. To your point earlier, I do empathize with the fact that it can feel like a black box because after I had boost posts, now what actually happens, right, And that's where the magic of the auction comes in. The auction is the heart

of Facebook's advertising system. That's how it figures out how much advertisers are willing to pay to reach someone like those hard to reach beekeepers we mentioned earlier. But where Facebook really excels is figuring out who those people are in the first place. So when we think about that, there's a few dimensions, and obviously it's a it's a

fairly complex system. But what we try to do is understand, Okay, what do we know about this person based on previous activity that they've had, previous ads, that they've interacted with um, And we consider also what is their likelihood of taking this particular action that the advertiser wants them to take, which is in this case actually Facebook uses stuff like age, gender,

and location, of course, but that's not all. I asked Ninkila what characteristics exactly Facebook was using to find these target audiences. How many signals are all in the I really couldn't tell you, because there's a whole bunch like liking, commenting on an ad, liking a page, liking a post, um or activities that you have done that advertisers are able to then come back and retarget um. And then there's also signals that we can extrapolate based on what

we know people like you tend to like. So if we know that you visited a particular retailer before, and there's another retailer that's selling a similar good, it's the likelihood of view interacting with that is also higher. So the answer to how many signals is I don't have a number. The software is so complicated that even people who work at Facebook can't see what's happening to the system as a whole. Knowing gets to know exactly who

is seeing a particular ad. So what about Facebook? This Facebook? Now, like if you go look at an advertiser, do you know exactly whose ads that person reached? Know? We we did not look at individual level data, nor do we surface it to your point, We do it at an aggregate level. We surface that to the advertiser at an aggregate level where we're telling them, but they're now at

an we don't go look at the individual level. Now. Yeah, So I guess, I mean, I guess it's one thing that's interesting about this whole system is it's it is so it's like, um, very automated and optimized. There are all these signals that go into it, and um, you know, out the other end comes something that's really useful and helpful. Um, but how do you how do you look for problems? Like how are you proactive about like the next iteration

of it. Yeah, that's a great question. We hold ourselves even internally, to very high standards when we think about how we developed, and it turns out with so much content and only so many people, rooting out problems is very very hard. Facebook doesn't really resp on two problems until they are reported by users, and sometimes it's people saying they're annoyed by seeing a certain ad, and other times it's much more serious when it comes to political advertising.

Obviously the stakes are high. There's still plenty of things that can and have gone wrong. That's despite the huge resources that Facebook is devoted to rooting out problems. There's just so much that goes unseen when you've tasked a smart set of computer programs with running your business. We've talked about the categories that get automatically created, like users who are saying into treason in Russia. Facebook had to go through and manually remove about five thousand of them.

That could be problematic, right, but they did so only after a series of articles from the news organization Pro Publica. It seems as though they're not proactively looking at the potential problems themselves, like taking action when they think it might be the right thing to do, like the day before the midterms, Facebook stopped letting Trump turn and anti

immigrant video into an ad on its network. It made the decision after major TV networks had already made it, and also the ad flip through its network in the first place. Also, once they decided to stop letting it be an ad, it was still an incredibly popular organic post on Facebook. Well, why change if you've got such

a successful thing going. What we saw with the Two Blind Brothers is that Facebook makes a lot of money off small, really small businesses that are targeting incredibly niche audiences. There's no other service, I mean, maybe Google's ad platform that can cater to such specific needs. The main area they've tried to tame is politics. A lesson spending just isn't a very big part of Facebook's business. The company earlier this year even considered eliminating and entirely given the

many headaches that generates for the company. But Sarah, the fixes generated headaches of their own, too, Right, You're totally right. I wrote this story a few weeks ago about how Facebook's new rules around political advertising now require anyone advertising about a sort of electoral issue to be verified, but the system was catching and taking down ads that hadn't gone through the verification process, and those ads sometimes had

nothing to do with politics. Well, so there was this vacation Bible school that got blocked because it was located in Clinton, Indiana. In an insurance company in Clinton, Iowa that had an ad about its family baseball night taken down because they had the word Clinton in their ads, as in former President Bill Clinton or Secretary of State Hillary Clinton. But it also happens to be a common

name of towns in the US. And I think we've heard so much about AI, but it's it sounds like Facebook came up with a logical solution to a need for more transparency and then signed an AI algorithm to do the work. But AI, despite the hype, it's not smart enough to figure out what actually might be political second,

making advertising work parents that people. Facebook is devoting huge resources towards making sure it's more proactive about political behavior on its site in I recently went back to Facebook headquarters and took a tour of what it's calling the

War Room. It's basically this room of a couple dozen people monitoring dat boards of articles that are trending and political ad money flowing so they can tackle problems related to election first one aunts so the biggest and most important thing we've done here in the last two years is huge advances and machine learning. That includes deleting fake accounts, sending fake news to fact checkers, and a lot of the issues that got it in trouble during the election.

Facebook has been showing this war room off to reporters as a sign that it's taking political content on the platform extra seriously. And they also have that at archive right where everyone can see who's bidding on specific ads. That's a big step, but there are still issues. For example, the feature that allows users to see who's paying for

each ad isn't perfect. A couple of Vice reporters realized that they could personally buy ads and claim that they were paid for by the Senate Republican Majority leader Mr. Racconnell. And in Virginia, a bunch of really savage attack ads ran against a House candidate, but the buyer's managed to completely obscure who they were or where the money came from.

In two days before election day, Facebook was alerted by the federal government about foreign accounts that could be trying to influence selection that it had to take down in the day after the elections, the company announced it was strengthening its verification process. But meanwhile it's still dealing with

a lot of influenced campaigns by foreign governments. So, Sarah, we've devoted this whole season of Decrypted to this idea of unintended consequences, and this is almost the biggest unintended consequence of all. Silicon Valley designed these systems to scale infinitely, very low human touch, um completely automated systems, and what they failed to recognize and which in this case had severe implications for the world, is that the problems scaled

right along with the systems. Right when you're prioritizing convenience and speed and low touch, right, these are systems where you really don't have to have a person selling and placing and add um, it makes it so easy to miss the little things. And what Facebook says over and over, you know, they point to examples like two blind Brothers.

They point to the many small businesses that Nkuila is helping out, and they say that the good that Facebook is done in helping these small businesses tell their stories and helping people have a voice in a way they wouldn't be able to afford to otherwise that outweighs the bad. Well, but to me, they just seem like different things. Like it was a system that was great for helping small

businesses reach big or even very tailored audiences. But when it comes to a tug of war over ideas and you know, and the balance between informing people and maybe deceiving them and to get them to do things they wouldn't otherwise do, that's where the system kind of breaks down. And I think you can't just go and say just because we've done a lot of good things means that

this service is good. I think, you know, so if you were to look at a government, for example, and say, well, the roads are really messed up, but the schools are great, Like, that's just not the kind of argument you would make about your government. You want all of it to work, and you want all of it to work in a way that's not dangerous for society. So let's take this to the to the logical extreme. And we touched on this earlier, like should Facebook just not allow a certain

kinds of advertising? Should it say we haven't figured this how to make this work for political advertising and not accept it? Well, then you get into the question of what political advertising actually is, right, is it something that is is just a candidate putting out an ad to get you to vote for them, or is it about these issues? And that the issues are really what Russia tried to stoke tension around in the election. They weren't

saying vote for Hillary, vote for Trump. They were saying you should get angry about immigration and refugees and race in America. And so I don't think that getting rid of a category of advertising really solves it. I think they have to think about how to proactively manage these systems with a little bit more human involvement, even if that doesn't scale. And verification obviously being so important, but it's been difficult for them to get that right. Why

why is it so? Why is it so hard? Well, when you're trying to verify, you have to go by what the people say that and they want it to be fast, and in politics, they've already gotten so many complaints about how slow that system has been, where people send in their verification and it takes a few days to pross us. Meanwhile, you know the rival candidate in a campaign is spending tons of money on their ads,

and that seems unfair. And so no matter what Facebook does, there's going to be this question of are they are they censoring or are they removing enough? And they have to figure out what the right balances, and maybe they'll never get there right and they can't be seen as favoring one side over another. And we should say this isn't just a Facebook problem, right, it's internet wide. It's an Internet problem, and it's a problem with this kind of this way that the assistance have been built over

the years. These scale so well, they work so efficiently, they're so high margin. Wall Street loves it. Uh, but you know, when it comes to the real world consequences, it doesn't work to just wait for users to report when things go wrong. So perhaps eventually there will be a system wide fixed. For now, though, I'm tempted to say that it's incumbent on users to just be more vigilant, to be smarter. Um. There, you know our our advertisers out there that are trying to convince you to buy

something new, maybe something you don't need. That's fairly innocuous, but in other cases they might be trying to get you to act in a way that you otherwise might not and that's it for this week's Decrypted. Thanks for listening. Do you have a story about advertising on Facebook? We want to hear from you. You can email us at Decrypted at bloomberg dot net or I'm on Twitter at Sarah Fryer and I'm at brad Stone. If you're a fan of the show, please take a moment to rate

and review us. It really helps new listeners find the show. This episode was produced by pa good Kari and Mage mus Hendrickson. Our story editor was Anne vander Nay. Thanks to Akito, Emily Buso, and Liz Smith. Friend Chessca Levie is head of Bloomberg Podcast. We'll see you next week. M