"Data Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Marcus Pratt, director of insights and technology at Mediasmith.
Since Facebook created its exchange, FBX, by opening up inventory to DSPs and other biddable display buyers, the industry has been full of chatter about the amazing opportunity this new medium presents.
But advertising on Facebook has been possible for years, so what’s really different between FBX and the ad products already offered by Facebook directly? Both options can provide advertiser value, though there are some key differences.
For the sake of simplicity, these options can be broken up into two categories: direct and FBX. Direct advertising runs via the Facebook marketplace, a sales representative or the Ads API, while FBX means buying through the Facebook exchange via a DSP or other FBX-approved platform.
In contrast, running through FBX is strictly a bring-your-own-data affair: All of the information Facebook has on its users is off-limits, but you can target people with your own first-party data. This may include retargeting visitors to your website or shopping cart abandoners, or any other cookie data, such as targeting “auto intenders,” as defined by a major third-party data provider.
The tracking capabilities may also vary significantly depending on the method used. With the direct option, performance data on impressions, clicks and frequency is locked up within Facebook, just as with targeting data. Advertisers can track users who click off Facebook to a destination page, but cannot always track view-through activity with their ad server of choice, missing activity from prospects that see an ad and don’t click but convert on an advertiser’s site. Larger clients and Facebook API partners have this capability, but many self-serve advertisers are limited to dropping a cookie only on those who click on their ad, not those who have viewed an ad.
With FBX, whoever is serving the ad can track impressions, clicks and conversions through their existing platform, such as a DSP or ad server, because FBX is treated like any other source of display inventory. This allows buyers to control and optimize Facebook advertising as part of a larger display-media buy, but information on performance by Facebook demographic segments is not available.
The initial popularity of FBX was largely focused on site retargeting. After all, the fact that so many people visit Facebook so frequently presents an enormous opportunity for an advertiser to reach more prospects faster, thus increasing the volume of a high-performing tactic. Prices are also significantly lower than display averages, potentially bringing greater efficiency than that of standard display.
Of course, media buyers looking to grow quickly sought to expand FBX use beyond retargeting. Some of these uses provide value; others are more questionable. Below are a few ideas media buyers have recommended:
• Targeting custom or lookalike audience segments: A high-performing audience-targeting approach -- which might target potential customers with similar characteristics as current high-spending customers, for instance -- could grow larger on FBX so it makes sense to add this inventory source.
• Overlaying standard third-party data segments: Buying data about purchase intenders or general demographic information to overlay on top of an FBX media buy can increase marketers’ reach against a specific target, but somethird-party data prices are high compared to the cost of media on FBX. Also note that some third-party audience segments can overlap with data offered by Facebook directly.
• Prospecting: This concept -- basically running ads on FBX without targeting -- can allow for high reach at a low cost but also raises the potential for significant waste. With all of the targeting capabilities provided by Facebook, why throw them all away?
For those not bringing unique data to the Facebook exchange, there seems to be one key reason to run ads on FBX rather than dealing directly with Facebook: view-through tracking. Since running directly with Facebook allows for click-based tracking only, advertisers looking at total responses will always see greater efficiency on FBX when taking view-through activity into account. Just because view-based conversion tracking is possible, however, does not mean FBX is more effective at driving true lift.
Consider the following example involving two advertisers with a $150 target cost per action:
• Advertiser A spends $100,000 on Facebook directly and tracks 100 click-based conversions ($1,000 CPA)
• Advertiser B spends $100,000 on the FBX and tracks 100 click-based conversions and 900 view-based conversions, for a total of 1,000 conversions ($100 CPA)
With this data, the first advertiser is not meeting his target CPA while the second advertiser is successful. However, just because the first advertiser can’t track view-through activity does not mean it isn’t happening. Conversely, just because the second advertiser tracks 900 view-through conversions doesn’t mean all were driven by Facebook. All advertisers will have a certain amount of “incidental” conversion traffic where an ad is served to someone who would have converted anyway. This can vary significantly by inventory source. Some suspect this level may be higher for Facebook because people visit Facebook frequently and often leave it as an open tab throughout their browsing experience, which makes Facebook one of the most likely last sites visited before a purchase.
FBX has clearly been a success for Facebook, with many advertisers jumping on board to retarget site visitors. I see an opportunity for Facebook to build on this success by combining the flexibility and trackability of FBX with the robust targeting it provides. Facebook undoubtedly has concerns that making more data more available would increase the risk of leakage, but it’s likely working to address this. The technology that Facebook acquired with the Atlas ad server could help.