After the Infinite Segment Model
It might have slipped past you in the news generally this week, or even in your more specific social media news, but the Facebook announced the removal of five-thousand (or more) possible targeting parameters and expanded advertiser education, a decision with important effects. Although a full list of removed targets was not presented, this does represent a major shift. Advertising Age noted that this was the long coming implication of Pro Publica’s investigative reporting on the abuse of ad targeting tools: Facebook’s segments could be used to ferret out private information or to deny housing or employment opportunities. This is a major step forward for the network that has only recently begun to take responsibility for the dynamics of the feed. What is striking is how this changes Facebook’s description of advertising, which until this point had relied on a vision of an ever-tightening targeting toward an infinite number of market segments.
The promise of Facebook for advertisers is two-fold: context and specificity. Communication theorist Raymond Williams described the core property of television, and I would say social media, as flow. Television was a perpetually moving medium, articulating one image to the next, with meanings blending over time. Commercials are the height of this logic of flow, without their articulation to proximal content, the business would utterly fail. Rules for such choices might defy reason: there are highly rated programs that would not be well suited for advertising. Negative campaign advertisements were not well tuned for inclusion during family dramas. The local news would be a better context. This logic of feeling rarely holds during election season and is true to a point, which is an important insight about media effects in general.
During the roadshow for the Initial Public Offering for Facebook, one of the featured products of Facebook was sponsored stories, which offered advertisers the opportunity to use social context (pictures of friends) to amplify the resonance of advertising. It is important to know which friend will help sell a cross-cut saw. For a number of good reasons, Facebook dropped this product shortly after the IPO. The promise was clear: good feelings to surround your advertisement will increase the likelihood of purchase. But unlike television, where good feelings would only appear between 8–9 eastern, an unlimited feed of goodness could be distilled. No longer would time during Perfect Strangers be scarce: the effect of the physical comedy of Mark Lynn Baker and Bronson Pinchot could be continuously synthesized from the abundance of inputs to the system. Beyond the raw volume in the system, many of the most affectively meaningful moments would be relevant for an extended period of time. New baby pictures remain in the feed far longer than pseudo-Polaroids of a recent lunch.
The threat to context comes from the corruption of the inputs. Just as the system benefited so handsomely from a positive cascade of new emotional support during the golden age of social media, a negative cascade caused by hateful trolls or the revelation of bad network management practices could cause the quality and quantity of the feedstock to decline. Working to make users feel safer and more supported by their feeds would be critical. Alex Jones might provide heat and light, but he sours the substrate. Damage in this area is not easily reversed, when users feel hurt or betrayed, the bad feelings linger. Rapid A/B testing, the core inductive logic of advertising is effective for finding which positive message will maximize sales. There is no inverse. Prevention by the ounce is worth more than pounds of cure.
In the past, market segmentation was a form of art: only a few segments could be selected and firms would hallucinate the features of the groups they sought to court. Firms came-up with characters that would fit with their organizational parameters, they give them personas like Rushing Rachel, attempting to tune their advertising and product to those particular imagined characters. This is a powerful labor saving technology. Deep research on the core customer could be focused and expensive advertising materials could be produced in more limited runs.
Facebook changed this by allowing highly specific segments like Episcopalian, tri-cyclists in Minneapolis. Facebook also lowers the standards for quality, users are looking for a basic text and a picture, not a multi-million-dollar production number. If you accept that targeting is the key to driving users into the funnel, a theoretical construct which sees conversion (a sale or vote) on the basis of a series of messages over time leading to a final message that induces the change infinite segments are the final, ultimate innovation.
Questioning segments is important. Every media student learns the key demographic, 18–49 year olds, and other relevant breakdowns of the population. An important challenge to the theory of segmentation comes in the reality that even well-advertised firms and segments fail. Advertising has an effect, but the idea of the funnel and the segment overstates the effect of the alignment of the segment with a particular text. Exposure effects are non-linear, diminishing returns are the norm. It is even possible that the last message in the funnel may be the least valuable, rather than the other way around. Proximity to the purchase means far less than the initial message that introduced a public to a brand.
Voted best read of Summer 2013 by the readers of Advertising Age, Byron Sharp’s How Brands Grow, posed the alternative to Facebook’s theory of the segment. If we understand the brand to exist across the entire set of possible segments, then it could be more effective to work on the development of better business process and basic identity. The hot trend is to recognize that better products and firms are more likely to succeed. Pressure on the dream of infinite segmentation and funnel development has been building for years.
What this means is end of the road for the theory of media effects that Facebook has promised to this point. Absent many, highly impactful demographic markers, it seems unlikely that the sort of inductive A/B testing that characterized the Trump campaign’s use of Facebook would be greatly dampened. The elimination of these segments then would deliver the social network giant into a position much more similar to a television network like CBS. Gone is the dream of asymptotically approaching the perfect fit of segment, context, and advertisement text. This is a profound and pro-social change that seems to be part of a larger theorization of communication by the company.
The magical assumption that drove Facebook had been that an ever-increasing din of voices could be effectively modeled for connection to particular advertising strategies. More speech would always be better, if you start from the assumption that there is a perfectly synched advertising strategy that pre-exists for every possible user and moment. Even more than the removal of particularly obnoxious users, the move away from the assumptions of segmentation represent a core change in the platform. Without the promise of infinite, perfect segmentation, the ethical balance between business and responsible platform management shifts, quality in context becomes suddenly, valuable.
