Working with Facebook Social Ads

So, here at Kinzin, we’ve been experimenting with Facebook’s Social Ads for the last few weeks, and I have a few results to report. Simple stuff first:

  • Click-through rates are abysmal. I was running the identical ad in about 15 different regions (you need to run them as separate ads to get the stats broken out), getting just over 10M views. Our average clickthrough rate was 0.06% (that’s 1 in 1513, for those counting at home). The best we did anywhere was 0.14%.
  • For some reason, we got quite different results (30-50% variance) if we ran exactly the same ad in exactly the same region, configured to show to men alone, women alone, and men and women together. For some reason, both together got much better results than either gender individually. Weird.
  • Again, the same ad: top four for clickthrough rates: Seattle, Portland, Alberta, and NYC. Bottom four: Toronto, the Maritimes, English Quebec, and Texas.

A little more subtle is the results from using “Social Actions”. That’s feature that Facebook advertises as being the differentiator for their ad platform. For those that don’t know what that is, Facebook will insert a blurb to let you know a Friend of yours has a relationship to the app or group that is the subject of the ad. A friend of mine might see, above an ad in the Facebook margins: “Michael Fergusson installed this app yesterday.” This is what Facebook has to say about it:

“What you’re looking at is a Social Ad. Advertisers provide the text, and Facebook pairs it with a relevant social action that your friend has taken. Social Ads mean advertisements become more interesting and more tailored to you and your friends. These respect all privacy rules; advertisers never have access to personal information about you or your friends….”

In theory, it could be useful to know that friends of yours use a particular app. In practice, it’s a bit creepy to see the name and photo of your friend in a banner ad. My advice in short: don’t use that feature. As I said in my last post, we’re still figuring out the rules of etiquette in this new space, but I don’t think Facebook (the company) has it quite right yet. For sure, it’s not right for us and our community.

On that note, I’d also like to apologize to any users of Kinzin applications that were creeped out by our (brief) use of that Facebook feature. We were as surprised as anyone by our negative reaction to seeing it in practice, and we turned it off as soon as we heard that others were feeling uncomfortable about it. We take our role in helping define this new space very seriously, and that role is not to push the boundaries of what’s acceptable, but to reflect the growing consensus of what’s desired.

It would be interesting to hear about other people’s experiences with the Facebook Social Ads. Anybody have anything to, um, add?

Privacy and Social Networks

ATM Privacy Area, by Cackhanded. CC via FlickrMark Ury has a great post on Privacy at his blog The Restless Mind. As I said in my comment on his Blog, I’d like to make an observation about why managing privacy (and other rules of social etiquette) is even harder than it seems.

Real social networks have actual humans as the end points in the graph. Complicated, technology independent humans. I have dozens, perhaps even hundreds of social networks I participate in, and each one has its own complex rules of etiquette and privacy, even when the membership of the network is mostly or even completely the same. In fact, it’s those rules that really define the network itself: the people I trust with my kids, the people I gossip with at work, or the group of cousins in my family that happen to be around the same age. Each of these is defined as much or more by what we do together (the “social grooming” as Robin Dunbar calls it), as by the membership, which may be mostly or even entirely the same. One reason for why these rules especially difficult to express in software is that these networks (especially the ones most established in my life) are typically multi-modal by nature. Take the network of “the people who love and care for my kids”, as an example: some are in FB, some are email-only, and some (like my Gramma) offline entirely. We humans are very typically very good at picking up on and managing these social “rules”, but often have difficulty migrating those rules to a new or unfamiliar modality of communication. As the number ways in which we can communicate with each other increases (more rapidly all the time, it seems), the harder it becomes to manage the complex social rules that govern human interactions.

Kinzin’s approach to this problem is to build what we call “Virtual Private Social Networks”. You decide on the rules and membership of the network, independent of the communications technology. This is obviously easier with smaller networks, and where the level of trust and familiarity is high, so that’s where we’ve focused ourselves. These Are My Kids lets a network of close friends and family share information about the family’s kids. The rules for privacy are set by the parents, and the invited members of the network can use (nearly) any medium they like to access the network: Facebook, email, postal mail, etc. This way, busy parents can spend their time thinking about what it is they want to say, and not worrying about how or where to say it.

Michael on TV


Pretty lame to post this a couple of weeks late, but I’ve been busy, then I was on vacation, then I was busy again, then I forgot… you know the drill. Anyways, I was interviewed on CBC television for a story on internet privacy. Here’s a link (thanks, Frank): http://gallery.mac.com/flee1#100129.

lab logoSticking with TV Fame, Episode 140 of The Lab With Leo Laporte on G4 Tech TV in Canada includes me doing a segment with Leo, talking about designing and building applications in the era of social networking. That interview will be made available on The Lab website after it airs on television.

My interview on Raincity Radio


Just a quick note that my interview with Dave O from Raincity Studios has made its way to the web. Since that interview, our membership numbers have nearly quadrupled, but the basic message remains the same. Those Raincity guys are a lot of fun – Dave and I bonded over our love of hockey history. He especially loved the vintage 1916 Vancouver Millionaires jersey I was wearing (see pic). Any other Cyclone Taylor fans out there?

Is the world normal?

Last week, 24Hrs Newspaper published a story about Kinzin’s Facebook plugin Are You Normal? in their print and online magazine (see the great photo on the left, by Rob Kruyt, that accompanied the article). In the article, I’m quoted as saying that we’ve reached 90,000 people around the world. Thanks to the power of the network effect, as of this writing we’ve already passed 220,000 (from 184 countries!) and still going strong.

The point of this post is not to toot my own horn (at least, not only to do that :-), but to mention for your interest that the next set of survey questions we publish will be written in part by our user base. We’ve had questions submitted from Finland, Spain, Greece, Australia, and the UK so far. I’m very excited about this in particular. As with many things, when we’re talking about what’s “normal”, what we choose to measure is often as interesting as the results.

I’ll keep you posted as things progress…

You can’t have small without the big


Robert Scoble, in response to comments made by Evan Williams, creator of Twitter, at Web 2.0 asked the question: “why would anybody want a social network with only 10 friends? Seriously, don’t we already have this? It’s called a family.” Williams was talking about creative constraints. Limiting the scope of a social network (TuDiabetes), or limiting the size of a message (Twitter/SMS). I’m surprised, frankly, that Scoble would make such an obtuse comment. That I have lots of friends isn’t a reason to not use Facebook, and that my family exists says nothing about the modalities of communication I use to “groom” my relationships with them (to borrow a term from Robin Dunbar).

TuDiabetes for example is a small network (recently passing 1100 enthusiastic users), constrained in scope to the issues related to living with Diabetes. That it is small and constrained makes it more valuable, not less.

On the opposite end of the scale, The Economist recently published an article called “Social Graph-iti – There’s less to Facebook and other social networks than meets the eye“. I think it’s intended to be a cautionary article about “irrational exuberance” in the social networking space, along the lines of the recent New York Times article. There’s a lot that’s right in the Economist’s article, but I disagree with a few things. I think the author doesn’t understand the nature of social networks in this respect: we can and do belong to many at one time (as we have since before there were “humans” at all). Many of our social networks, in fact, are built on top of other, existing networks. For example, the management team I work with at Kinzin is a small network built out of my larger “business associates” network, which is part of my “everybody I know” network. It overlaps with my “close friends network” and my “co-workers” network, and so forth.

But unlike other networks, social networks lose value once they go beyond a certain size. “The value of a social network is defined not only by who’s on it, but by who’s excluded,” says Paul Saffo, a Silicon Valley forecaster. Despite their name, therefore, they do not benefit from the network effect.

Mr. Saffo is both sucking and blowing, though, using “Social Network” to mean two things at the same time: the sum of all the users who are members of a particular social networking application, and all of the connections that each individual member has on that network.

My personal network doesn’t scale for all the reasons that I’ve written about previously in my commentary on Dunbar’s Number , but Facebook’s network doesn’t have the problem in the same way. It can contain every human in the world, and it doesn’t lose value for me (at least not the same rate or in the same way), because I only need it to be able to find everyone in the world who potentially could be my contact, I don’t need everyone to actually be my contact. It seems to me the Economist doesn’t account for the fact that we all have multiple overlapping networks, containing people that in the end all are drawn from the same pool, namely all the humans in the world. If I’m trying to maintain 10 different social networks (friends, family, business, acquaintances, etc.), it’s helpful to have the underlying system contain all the people in all the networks. So Facebook having 6B members helps me to better create the 10 person network that represents my geographically dispersed family.

Everyone’s trying to make it simple: “Irrational Exuberance!”, or “Everything’s Really Different!” The reality, as it usually is with humans, is much more subtle.

What’s "Normal" anyway?

The Normal Theatre, by K2D2vaca

Our Facebook adventure sure has been interesting. In the week since we launched, we’ve had over 15,000 people do our surveys (Update: three days later, and we’re now over 22,000…) and discover just how (ab)normal they are. One curious thing I’ve noticed while discussing “Are You Normal?” with people is that, at least among the people I talk to, most people assume that their normalcy rating will be very low. In fact, being “abnormal” seems to be what they’re hoping for. The thing is, the system only calculates your rating based on what everybody else said, so if everybody’s a bit strange, well… that’s what’s normal. It’s what I really like about this application – the community decides what’s normal, not us. We could have used some standard psychological test and given a stock answer, but everybody deserved to be judged by a jury of their peers, don’t you think?

In case you’re wondering: I’m 23% normal (and falling).

Normal, by Binderboy

Which brings up the other interesting side-effect of the way we calculate the answers: that your rating can and does change over time. As more people answer, the most common set of answers changes slightly, effecting your rating against that “standard”. To take advantage of this interesting side-effect of our rating system, a new feature we’re planning is the ability to check your rating against specific groups – your own friends, for example. And when Facebook launches their new “contact grouping” feature, you may be able to compare yourself against particular sets of people – work, family, whatever. Let me know if you think this feature would be really interesting to you – if enough people call for it, I’ll get the development team to move it up the schedule.

Some tidbits, gleaned from the results so far:

  • 68% of people answering the surveys are very concerned about the environment, or are taking action to do something about climate change. 9% say they’re not concerned, and a full 23% don’t take either position, which is interesting.
  • 9% describe themselves as conservative, 25% as liberal and the rest (65% or so) describe themselves as non-partisan or none of the above.
  • 25% think that a family should have only a mommy and a daddy.
  • 41% of parents lied, saying that having children hasn’t effected their sex life ;-), the rest need to get away for the weekend.
  • 45% wish that their kids knew more about their family history and culture

Stomen
There will be a new survey in the next day or so (Are you a normal Facebook user?), and some UI improvements, so stay tuned.

Cross-posted at the Kinzin Blog.