By diversifying across a portfolio of Facebook audiences, an advertiser can promote maximum reach, achieve optimal pricing, and access the full bounty of audience intelligence.
A simple and straightforward way to achieve audience diversification is by segmenting the campaign portfolio among different age groups. More than 95 percent of the U.S. audience on Facebook shares an explicitly targetable age, so with age it is very easy to create a set of large and non-overlapping audiences.
However, by using age to drive audience segmentation, advertisers ignore some of the most effective strategies for Facebook campaign design that we have discussed elsewhere; by targeting likes and interests, advertisers can significantly increase the odds and the magnitude of a campaign’s success.
Navigating through Facebook’s gigantic and shifting universe of likes and interests can make a sandstorm in the Sahara seem comfortable in comparison.
Liking Begets More Liking
But if there is one clean and pithy truth to be extracted from Facebook’s web of likes, it is that liking begets more liking.
Given any common interest in Facebook’s universe, we estimate that users with that interest over index for 94 percent of all other interests. Liking Facebook objects turns out to be a sticky behavior, which poses difficulty for those trying to builda portfolio of diverse audiences. The same eggs are in the same baskets.
An online news property that has had success running an awareness campaign to fans of NPR, for example, might decide to scale up its approach by targeting fans of PBS as well. While this is an intuitive strategy, to do so in the spirit of diversification can be slippery: Almost 60 percent of the Facebook audiences for PBS and NPR overlap. This can be ineffective for three reasons:
- scale suffers, because 60 percent of the ads are being delivered to users targeted by both campaigns;
- pricing is suboptimal as in some cases the advertiser is bidding against itself for the same impressions; and
- there is less that is ultimately tested in this scenario and therefore less to ultimately learn about a brand and an audience.
Similar phenomena can surface outside of the likes and interests domain, especially with Facebook’s fluid, multi-tiered options for geographic targeting. For instance, where does the 15-mile radius around Minneapolis end, and the 25-mile radius around St. Paul begin?
In order to promote maximum reach and optimal pricing, and capture the full bounty of experimentation and discovery, we are taking a portfolio approach to the design and execution of Facebook campaigns.
Layer The Data
Such an approach may include the layering of census data on top of a retailer’s own proprietary data on store locations and sales, to create a perfectly disjoint, evenly sized, and locally aware series of online-to-offline campaigns.
It may involve allocating a gaming company’s carefully manicured, long-tail list of keywords into minimally overlapping and thematically distinct buckets, or targeting a business-to-business campaign to sets of non-contiguous workplaces categorized by sector.
In any case, the goals are always the same: maximum reach and optimal pricing, and the full bounty of experimentation and discovery.
Guest writer Rob Leathern is chief executive officer of optim.al.