Optim.al: Accessing the Facebook API

Facebook is huge. It has over half a billion sign ups and even allowing for dummy and secondary accounts that still leaves a lot of people actively engaged with its service. A good deal of Facebook’s potential commercial power is based on the access to personal information that it has obtained from its user profiles and the monitoring of their activities, their likes, their shares, their posts and so on. But even an enormous, well-funded company like Facebook which has the ability throw immense resources at any given challenge realises that it cannot do everything itself.

As Sasha Ziman, Sales Director of Optim.al, puts it, “The notion of the API and the development of relationships with companies such as Optim.al is that they know they don’t have all the answers and so they have invited vendors to try and innovate using their tools.”

Optim.al is a platform where advertisers, big and small, can go on and create campaigns and test a lot of their creative ideas for their Facebook campaign and figure out which creative ideas and variables are the best.

Optim.al is a part of Xa.net which was founded by Rob Leathern who in his own words, “engineered a great deal of multivariate testing and optimization for display ads, emails and text ads at LinkedIn and NexTag and saw an opportunity to extend these methodologies to the rapidly growing and evolving Facebook ad ecosystem.”

Optim.al are among the very few Facebook data API partners and they have direct access to the Facebook ad-server. Their multivariant platform tool can substitute for Facebook’s own interface.

The advertiser can give permission to Optim.al to target their connections so they can advertise to them. Or, very neatly, they can advertise to everyone except their connections. They can also reach out to the friends of their connections which for many Facebook users can be a very large group.

The advertisements themselves are broken up into the headline, the body copy, and an image. The primary body text is broken up into two pieces. It was felt that the ‘call to action’ should be dealt with as a sole element and is therefore treated as a variable itself.

Sasha explains further, “[With this creative optimization platform] not only can we test out which ads and parts of ads are doing best but we can also figure out which audience components are successful as well. Not only can you get indications of responses from gender you can also break up the analytics according to their likes and interests.”

The full factorial of options, the multiple and varied ways that and advertisement can be created, combined and recombined can get deep, complicated and unwieldy very quickly. The issue with running every possible combination of an ad is the time and energy that is required to sift the information from the noise.

Fortunately, to help users avoid being overwhelmed there are simpler routes through the system. There are videos available to help with working through the options and it is possible to outsource the entire experimentation to Optim.al itself.

There are a lot of the other platforms which help advertisers manage their Facebook campaigns but this technology which was developed in-house by their own engineers has the ability to give its users insights to improve campaigns that they might otherwise have found very hard to obtain. It enables advertisers to pinpoint what makes their creative work great or not great.

The key here is audience segmentation, audience targeting and really creating ads that resonate with that audience. The power of being able to access a service like Facebook directly is that you have hundreds of millions of people who will tell you exactly who they are and what they like.

Whether Facebook will open its doors further to allow more third parties to have access to their API has yet to be seen. Considering how valuable the social graphs are that it possesses it us unlikely we will see a third-party eco-system that exists around say, Twitter.

However, the willingness of Facebook to allow companies to have direct access to its ad-servers may be indicative of an increasing self-awareness of the limits of its own capability that comes with maturity borne of experience.

Starting Out With Facebook Targeted Advertising

I recently ran my first Facebook advertising campaign in order to promote an engineering degree course here at NUI Galway. Although it may have been run too late to change most students’ minds about what course they wanted to do, I thought it might be interesting to share some insights into the Facebook advertising process, and to give an idea of the power of the targeted advertising system they provide.

When you click on the “Create an Ad” button on Facebook, you are presented with some ad design options – namely, the URL where you want people to click through to, the title (limited to 25 characters, which requires some imaginative juggling of words and txtspk), an image, and a description for the ad (135 characters). This is standard stuff, but where it gets really interesting is when you start playing with the targeting options.

By default, you normally need to choose a target country. Choosing Ireland will inform you there is an estimated reach of 1.5 million people who are 18 or over on Facebook. Choosing the USA changes that figure to 120 million. Considering the population estimates for Ireland and the USA are around 4.5 million and 300 million respectively, that’s at least a third of each country who use the site. You can choose countrywide or city-specific targeting.

You can then fine tune in terms of demographics: age and sex. But clicking on the less obvious “Advanced Targeting Options” link shows some fascinating options: target people on their birthdays; target people interested in men or women; target people in a particular type of relationship (single, engaged, married, other); and languages. You can also target people at different education levels and in specified workplaces.

Finally you choose your daily budget, and also whether you want to go for eyeballs on ads (cost per mille – CPM) or clickthroughs (cost per click – CPC), along with a bid amount for how much you are willing to pay for impressions or clicks.

This may sound like a bit of a black art, and since it was my first campaign, I decided to adopt a gambling strategy by placing my money on different horses. I knew that parents were highly influential in their child’s choice of college course, but I also knew that I primarily wanted to show the ads to school-goers. Then I wanted to make sure that I covered other groups like teachers, relatives, etc.

So I went for a three-pronged approach:

Secondary school students
…exactly between the ages of 16 and 19 inclusive
…who live in Ireland
…who live within 50 miles of Athlone, Ballina, and about 30 other towns and cities in the west of Ireland
…selected cost per click (CPC) was 40 cents

…exactly between the ages of 40 and 55 inclusive – (I was estimating an age range of the teenager’s age plus 24 to 36)
…who are married – (I know this often may not be true, but it helped cut out many wasted ads to non-parents)
…who live in Ireland
…who live within 50 miles of Athlone, Ballina, and about 30 other towns
…selected cost per click (CPC) was 60 cents

Everyone else
…who live in Ireland
…selected cost per 1000 impressions (CPM) was 20 cents

And here are the results. For about €520, my ad was shown about 7 million times, and was clicked on about 1,750 times. That’s an overall average cost of about 30 cents per click.

It’s difficult to estimate if a click is worth more than a printed flyer sent in the post, but if you know roughly what type of clicker you had, then it’s pretty valuable information. Of those 1,750 clickthroughs, 1,100 were from the “secondary school students” group. 250 were from the “parents” group. I had budgeted about twice as much money for students, but for a lower bid rate they actually got double the clickthrough rate of the parents. The “everyone else” category was shown as much as the other two campaigns put together, perhaps due to the wider geographic spread, but received less clicks than the students, perhaps due to the ad placement (CPC being prioritised over CPM).

Clickthroughs for the three campaigns. The dip is due to my credit card maxing out while travelling!

What was interesting was that the most successful campaign was indeed the targeted-to-students one. But I was surprised that the broad spectrum campaign outdid (by 150 clicks) that of the one targeted to parents in the region, for the same amount of money spent (€130 each).

Facebook offers very fine-grained reports on campaigns. There are also stats regarding “social clicks”, i.e. the number of ad impressions where the viewer saw that a friend had liked the ad. Interesting stuff, and it seems there is still lots for me to learn…