So let's unpack the Facebook Pixel a little bit so that you can understand it better conceptually. I love to show the browser extension, Ghostery when I give these examples. Because really all it is is a little tool that you can activate from the top tool bar of your browser. And when you're on a website and you engage the little Ghostery tool here, you actually can see all the different trackers that exist on that website. And so the the Facebook network is essentially one of those trackers and in fact here, we don't see Facebook. So for the New York Times website Facebook isn't able to influence the behaviors that are happening on this actual news site. It doesn't participate with the Facebook tracker so that data isn't available. What is available here is DoubleClick, which is owned by Google and Amazon among others. So again, what's happening here? We're visiting the New York Times and we're seeing that there are three major ad trackers that track the behavior of me on this website. So now how do they know it's me? Each one of these advertising trackers has an ID that's associated with you. It could be associated with your browser or it could be associated with for Facebook's instance, your Facebook account, but that ID follows you around the web. And as you visit websites that ID links back your behavior to your account, so things can be added up one to one. So let's unpack another example. Here, I am on buzzfeed.com and you can see that there are four trackers associated with it. I can see that one of those trackers is what is noted as Facebook custom audience. So this means that BuzzFeed has basically put a little snippet of code on this site that allows Facebook to listen in on this website visit. So Facebook now has this data and Facebook now knows that I visited buzzfeed.com. Similarly BuzzFeed now can actually use this and say hey, I'd like to send a Facebook ad to everyone that's visited my website in the last month and send them a article that I think they might like so that I can get them back on the website. This is kind of the ebb and flow or the give and take of this type of relationship and why it exists on so many different popular websites. Facebook is not only collecting the data on the consumer and gaining knowledge, but it's passing on some knowledge back to BuzzFeed in exchange for the data. So this Facebook Pixel tracking phenomenon isn't just limited to general browsing behaviors, but instead actual companies can track specific behaviors. So if I'm a coffee shop, and I want to sell bags of coffee on my website, I actually have the ability to add a specific behavior when someone adds a bag of coffee to their shopping cart. Let's consider the scenario where consumer adds a bag of coffee to their cart, but then has cold feet and doesn't check out. With the Facebook Pixel if we register at that moment that somebody added that bag to their cart and then notice that they didn't complete the follow-up action, which was then actually buying that bag of coffee. We can consider these people as very, very close to converting. And with Facebook Pixel we can easily say I would like to serve an ad to folks that just added a bag of coffee to their cart but didn't check out. This is the ability that Facebook Pixel adds to what we call retargeting or remarketing. This is the ability to go after consumers that we've already had an experience with. Maybe they don't even have an account registered with us yet. We don't even really know their actual name, all we know is that their browser or their device almost bought a bag of coffee. Now, let's try to reach them on Facebook and try to get them to convert and finish buying that bag of coffee. It gives an Amazon example. Let's say that I was researching smart thermostats and I wasn't sure about which one to buy, but I put a few in my cart just to consider them so that I could come back later and make a decision. Amazon now knows that I'm interested in smart thermostats. So don't be surprised if you see an ad on Facebook later that day encouraging you to buy that thermostat. This is a perfect product example because there's enough margin here in the sale, Amazon makes enough profit so that it can afford to serve you this ad. So Facebook actually gives you the ability to look at your advertising data for you and your account, and I really encourage you to do this. It's under the Privacy Settings of your existing Facebook account. All you have to do is go log into Facebook and then click on Privacy and then you can click on Ads, and you can actually get access to a lot of these pages that I'm about to show you. So Facebook actually shows you now who has uploaded a piece of data about you. So each company that you see here represents a time in which data was uploaded that matched my Facebook account. So let's use an example a kind of pack what that means. JetBlue here is an example of an airline that I have flown, but that I don't fly at all regularly. So JetBlue has my name, my birthday, my email address, my phone number, tons of data about me. Really, in general all you need to connect to consumer to their actual Facebook account is either their phone number or their email address, so they have plenty. JetBlue has uploaded my data to Facebook in the efforts of trying to be able to advertise to me on the platform. So it uploaded my data in the form of a customer database and it said here are people that regularly fly JetBlue or maybe even here are people that don't fly JetBlue often. These are people that we'd like to reach with advertising to try to convince them to fly JetBlue and more, or to tell them about sales, or to tell them about a promotion that were running. This custom audiences feature gives you the ability to take your customer database and actually put it on to Facebook so that you can reach these customers on the platform and advertise to them. It's really connecting you to your consumer through Facebook. And this is why so many companies do it. So who has an email database about me? Pretty much anyone that regularly emails me or anyone that I bought a product through online, there's an easy to access database. Similarly square and other point of sales transactions devices also log email addresses. So it may be possible for you to think about a custom audience data set from an existing database you have such as your point of sales device. So I'm here in the Facebook Ads privacy section, and I'm looking at my interests. And I noticed that the people that I'm most interested in are all musicians with the exception of one. So now I'm beginning to wonder where is Facebook getting this data, and it's actually pretty simple. Where do you think Facebook has the best likelihood of knowing what musicians I like? Well, that would be through Spotify. So Spotify actually has the ability to know all of my musical interests in all the people that are associated with it. Facebook is taking this data because my accounts are linked in essentially connecting the inference and saying if he listens to this music often, he's also interested in these people. In Facebook interests are really a mixed bag. Here, we're looking at the Sports and Outdoors category and it really doesn't match up that well to my true interests. I have no interests in Maryland football or Philadelphia Phillies. I barely know how to ski. I'm not a great soccer player and I'm pretty bad at basketball. So these general interests here don't map very well or the precision on them is less than 50%. So again, when you're building targeting parameters that leverage these interests you're going to have to try many because not all of them will be highly accurate. And then finally, the lifestyle and cultures interests tab here is just completely off, just further reinforcing that not all interests are accurately influenced on the Facebook platform. So why can inferences be inaccurate? There's a number of reasons most common is just somebody else using your device or phone. So if somebody Google something from your computer often that can be registered as become as coming from you. Facebook doesn't know the difference between you using your browser and somebody else using your browser. And often people Google things that they're not really interested in but they maybe just have to look up once. Some people may only be interested in mattresses for an hour or two before they make a purchase. Facebook will register being interested in a mattress is something that should last for weeks. So if you look at one web page your interests may follow you around even though they're not your true interests. Pixel data can of course, just be wrong anything with the scale of tracking millions of website visits a day has errors and bad data is collected at times, so that could just be a flaw in the system. And then finally consumers just change a lot too. So there's not really a guaranteed way to know that just because someone was interested in something once, means there are going to be interested in it for a long time. And that really is some of the issues that we see with my interests, maybe even some of them were old interest from many years ago, but they've now changed as I have changed. I have to say the one space in which I was most impressed by the inferences that Facebook has made about me was in my category section of this Facebook ad data. It really nails a number of really important things about me and where I am in my life. I'm shocked that it knows that I'm away from my family. I mean that takes a number of pieces of data to put together. It has to look at my relatives and notice that my relatives live thousands of miles away, but it is correct. I'm also impressed that it seems to somehow know that I'm away from my hometown. I've long since removed those items from my profile, yet it still has a way to inference that a large portion of my friends live far away.