The first thing you’ll find when you google for “Facebook Ads” is, besides the official Facebook page, a bunch of guides that will explain you how to make good use of them. Because the people who write them probably started using Facebook Ads after years using Google Ads, the main message in most of them is: Facebook Ads are different from Google Ads. The most obvious conclusion is that, instead of putting your links next to search results for “tents”, you need to target people who “like” camping.
A more interesting side of the story is the structural difference. Ads on Google have higher click-through rates, because they are about something specific the user was searching and was interested in even before he opened the browser to look for them. Facebook Ads, on the other hand, are a little bit like billboards on the side of the road. You notice them, but you’re more interested in your friends getting married or throwing birthday parties. You don’t necessarily pay much attention to the Ads consciously, but you look at them for quite a long time (while Google search results are gone after few seconds).
Up to now, Google has had limited success attracting branding campaigns on its search results. Try to google “cool jeans” and, believe it or not, you won’t see many major brands among the ads on the right side of the screen. The ones you do see are usually advertising a discounted sale, not how amazing their products are. Retailers take the lion’s share of the ads. I’m not a marketing expert, but my guess is that they assume that once you are at the googling stage, either you have already decided and want to buy online or you’re looking to compare prices and characteristics. Brands are not very interested in trying to change your perception of them at that stage, because the event of “googling for a given product” is too infrequent to influence your opinion significantly and happens too late in the decision process.
On the other hand, seeing a brand multiple times, while you’re thinking of something else, is exactly what it takes to make you familiar with it and make you think it is cool. A roadside billboard you see every day on your way to the office does that job.
Facebook Ads are more like billboards, but with some advantages:
- Users can click on them and see the brand website
- They are surgically targeted to the people you want to reach
- Lately, Facebook introduced the “Sponsored Story”. If a friend of yours “likes” a brand on Facebook, you probably get to see it once on your feed (if at all). Sponsored stories give brands the opportunity to amplify that event and show it to you multiple times, on the right side of the page, together with the ads. For the past few months, we’ve kept seeing that our friends like Dior, Calvin Klein, Tommy Hilfiger, etc… and if our cool friends like a brand, we’ll obviously buy it too, won’t we? I can’t help admiring the idea of Sponsored Stories. Word of mouth (people telling others that they like a product/service) has always been extremely powerful. What our trusted friends think (and, to a smaller extent, what anybody around us thinks) strongly influences our perception of a brand, to the point that some brands actually faked word of mouth, by paying people to have enthusiastic and loud conversations in buses or subways about a given product or service. Facebook Sponsored Stories are more subtle and effective: they do not fake word-of-mouth, they amplify the real one, coming from our friends
Could Facebook Ads be even better than they are?
In a word, Facebook Ads are quite advanced. Could they be even better? I believe they could. I’ll focus on the second of the three points above. At present, Facebook allows advertisers to choose target audience by location, demographics (age, sex language), marital status, likes and interests, education and work (this is true for “normal” advertisers – key accounts might enjoy a little more privileges, but I don’t really know). While this looks like a dream to a marketing manager who spent his life buying space on TV and magazines, this is surprisingly little compared to how much Facebook knows about us. Facebook has the opportunity to be a lot more granular in segmenting the audience. Here are some ideas:
- Depending on the IPs I log in from, Facebook knows how much I travel. This could be extremely useful for a telecom company raising awareness for a roaming promotion or for a credit-card company trying to increase cross-boarder transactions
- Facebook has a pretty accurate estimate of how much money I make, from many different sources (how much I travel, the addresses of the hotels I log in from when I travel, the addresses of the houses I log-in from, the company I work for, etc…)
- Facebook even knows how beautiful I am: here’s how. When you look at a friend’s wall to write happy birthday to her/him, on the left you see some of her/his friends. If one of them is particularly beautiful and you’ve got some time to waste, even if you don’t know her/him, you might click on her/his photo to see it full size rather than just the “friend thumbnail”. Now, Facebook knows how many people who don’t know me directly have seen my face by chance in the “friends” section of a common friend’s profile. Facebook could track the share of these people who clicked on my face and I bet that there is a high correlation between my beauty and that share. Wouldn’t this be useful for a fashion company offering discounts on trendy clothes? Or for a model agency’s recruiting campaign? (I’m writing it as if it was a breeze – obviously that ratio is just the beginning of the development of a beauty score, which would need to be sophisticated and well tested)
- Facebook knows how smart I am, because of all the IQ test games and application that are on Facebook (I’m referring to the part of intelligence that is measured by logic tests). This would be useful, for examples, for companies recruiting people for roles where performance is correlated with analytical skills and IQ
- If my friends comment on my wall more often than they comment on other people’s walls, then Facebook knows that I am popular among my friends and that I’m likely to be good at entertaining people. This could be used by companies trying to set a new trend or promoting a new club: if I am popular and I propose to go there, my friends will follow me
- From the photos I upload, Facebook knows what activities I enjoy more than just my likes and interests say
I’m obviously not saying that it would be easy to put the above into practice and to make money thanks to it. Even assuming that it is possible to get reliable estimates of the beauty, intelligence, popularity, wealth, interests and traveling habits of users, bringing them to the market would be even more difficult.
On the technical side, adding more variables to the available segmentation criteria might make the “cells” too small for the auction to be liquid (ads prices are determined by an auction: for a given segment of users, the price is the “cut-off” point in the supply-curve that fills in all the available ads on the users in that cell. If the cell is too small, this process might become problematic). On the commercial side, making advertisers trust such sophisticated segmentation (so that they are prepared to push the auction price for the very sought-after micro-segments) would take time and effort.
Anyway, on the technical side, math could help. Multi-linear regression helps insurance companies price their products with tens of millions of different “pricing cells” even if the actual number of customers is just a few hundred thousands. It could also help attribute a fair price to “cells” that do not contain enough users to run a point-wise auction. On the commercial side, it would probably be possible to identify early adopters and create success stories to convince the rest of the clients.
I’m curious to see what will be the future of targeting in on-line advertising. What do you think will happen? I’d be happy to read your opinions in the comments to this post.