Digital Advertising, Peer Networks, and Revealed Preferences
The advertising world has been disrupted by digital technologies. The digital advertising space has been growing by leaps and bounds as traditional advertising channels in print, tv, and radio have dissipated. Industry analysts anticipate that in 2018 online digital ad spending will surpass traditional in market share.[i] Of that digital ad market, two firms — Facebook and Google — have captured two-thirds share. Madmen truly no longer work on Madison Avenue.
But this disruption is far from over, and that includes the future of the duopoly.
For an economist, advertising resembles an odd duck. Economists deal primarily with revealed preferences of consumers and other actors: individual preferences are revealed through concrete actions such as sales or, in the case of politics, votes. But advertising mostly operates in the realm of what we might call implied preferences. Implied preferences are based on hopes, desires, opinions, etc. This is what advertising seeks to shape and discern in order to convert to actual revealed preferences. To an economist all this looks like so much smoke and mirrors.
In politics and public opinion, the difference is often referred to as soft vs. hard data. Opinions and surveys are soft data; actual votes are hard data. Advertising is a major part of political campaigns because nobody really knows what the soft data infers until after the hard data has determined the results on election day, so, naturally, a candidate or political party desperately wants to shape that soft data. Since data populations are often extremely large and immeasurable, sampling is also used from both soft and hard data to make these inferences, increasing the margin of error. Lastly, people can act strategically instead of sincerely revealing their true preferences. The take-away here is that there is a lot of uncertainty in the science and art of advertising, marketing, and promotion. This uncertainty is the crux of marketing analysis and where the data miners come in.
There is a mathematical proof in statistical theory that more data gets us closer to reality, or the truth. In statistical terms this can be simply stated that as population or sample size N grows, the margin of error for inference decreases. So more data gets us closer to the truth. This is what large data platforms like Google and Facebook bank on. The point of difference between the two is that Google relies on data gleaned from search algorithms to infer consumer preferences, while Facebook relies on data sharing among like-minded peer networks to inform those preferences.
We may search for new ski gear on Google, or a friend on Facebook may post information about cool new ski equipment he has recently used or bought. The question for advertisers is which means of invoking our latent preferences is more likely to lead to an actual purchase? The other question is whether advertisers can reach and shape the preferences of customers who don’t yet know what they want? Between the two platforms, they place their bets accordingly. In either case, the first thing advertisers must do is capture their intended audience’s attention — a scarce commodity amid all the noise in the digital world.[ii]
In the comparison between Google and Facebook, Google seems to operate closer to revealed preferences because users actually act intentionally through their search routines. Google extrapolates search data and auctions off key search words they call Adwords. Google then propagates paid advertising on our specific search and content pages. For our purposes here, Facebook presents the more interesting puzzle because preferences are inferred or implied by the connections of peer networks.
All social networks, like Facebook, are peer networks; we may call them friends, family, colleagues, contacts, or followers. So an online social network (OSN) platform is a web constructed of many overlapping peer networks. To discover the viability and value of an OSN in terms of solving the advertising puzzle, we need to evaluate the data sharing that connects these peer networks. In other words, do we really adopt the behaviors of our friends and family? Are 600+ “Friends” on Facebook really our friends? What does 500 “Likes” really mean? Do we really act on the messages we receive from those influencers we follow on Facebook or Instagram?
Two relevant variables that can help us answer these questions are trust and commitment. Do we have a high degree of trust in the data that flows through our connections? Are we committed to those peer connections in some way? The recent experiences of Facebook reveal how difficult and precarious it is to measure these variables. In many cases, trust and commitment are subjective and open to manipulation. The issue crosses over into that of individual privacy and security and whether two parties establishing trust and commitment want the nature of that relationship revealed to third parties.
Possible solutions to these problems are being addressed with new technologies like blockchain distributed applications and tokenization. These efforts focus on reducing the uncertainty inherent to information asymmetries by either eliminating those uncertainties (with public blockchain ledgers) or objectifying the subjective (by curating with incentivized voting mechanisms using crypto-currencies or tokens). In a certain sense, blockchain seeks to offer a “trustless” transaction platform, while tokenization seeks to incentivize and operationalize invested commitment and data prioritization. For example, token curated registries (TCRs) are an attempt to filter distributed information by curating the prioritization of lists — much like how The New York Times or Billboard produce lists of the Top Ten songs and Best Sellers book lists.
The challenge has not yet been met but solving these problems offers the promise of creating new value to the economy and reducing the waste of inefficiency costs, especially that of advertising, marketing, and promotion. An economist would be cautiously optimistic, an ad exec or lawyer perhaps not so much.