Facebook Age Calls for New Metrics

Data-Driven Thinking"Data Driven Thinking" is written by members of the media community and containing fresh ideas on the digital revolution in media.

Today's column is written by Nikhil Sethi, who is co-founder/CEO Adaptly.

In the world of social media and social networks, native advertising models are taking over. Anything I can do as a consumer - tweet, like, share, stumble, watch a video, comment - is reflected in the possible paid media opportunities. If I as a user can tweet, the paid media opportunity is a promoted tweet. If I can like something or have a conversation on Facebook, the paid media opportunities lie within sponsored stories and page post ads. The traditional display advertising approach of using real-estate to push content into users' view is starting to erode very quickly and the use of data in marketing starts to shift from, "Give me a vacuum and I'll go sell it" to "Let's figure out which vacuum to actually build."

The recent news that GM is claiming Facebook ads are ineffective actually makes a lot of sense from the traditional point of view. Of course Facebook is unable to compete - if the measure of effectiveness is a legacy attribution model. We're comparing two fundamentally different approaches to each other. Intent-based advertising models are akin to the demand harvesting approach, which Google has conditioned the marketer for several years now. There are no well-respected models for a demand generation approach in the digital landscape today. Radio, print and television have adopted metrics that represent top of the funnel campaign goals but these don't translate gracefully into the digital space.

Traditional methods for measuring the quality of ads in search and display channels have focused on contextual relevance and click behavior. These methods focus on the first order impact of an ad unit and the site it appears on. But the powerful effect of the democratized social web is that a single voice can change perception around a brand or other social object extremely quickly and virally. We should be measuring the impact of a piece of social content not by how many people have clicked on it, but by how it organically travels throughout networks. Because of the way content moves within social channels, we should consider determining ad quality in the format of percent penetration (organic reach) within a channel. High penetration shows that a given audience resonates with a piece of content, whereas low penetration indicates either a poorly acquired audience, poor content being generated or weak potential for sharing.

Social platforms are having an identity crisis. Facebook looks, feels, and smells like a digital platform, but it isn't. The cookie-tracking model was version 1.0 of identity on the Internet and, simply put, Facebook is the next generation version of identity.

The growing expenditure on the earned and owned content side of social advertising is part of this new model. Online advertising cannot be solely focused on demand harvesting or responding to the last keyword used or page viewed, but should also be about generating demand, pushing people towards fresh new content and ideas and spreading them further with paid media.

In a social world where everyone's voice is equally loud, paid media is only as good as the content behind it, and content is only as good as how many people it reaches. The fundamental measure of effectiveness in advertising has evolved. We need new models for the more social web.

Follow Nikhil Sethi (@nsethi), Adaptly (@adaptly) and AdExchanger.com (@adexchanger) on Twitter.


  1. Alex Andreyev

    Great piece. Completely agree. The problem is that the advertisers are looking to streamline a process of media push, assembly line of media based on version 1.0 of internet identity. Search, display, video, mobile, social is all biddable, thus it fits the assembly line model, however this is only from a business standpoint. It doesn't fit a typical consumer model of purchase behavior, which is why one stop "DSP" across all media for buying won't work (although I think it will work magic for analytics). So rather than building shops of optimizers and executioners, advertisers/agencies need to start investing into strategy specialists for each media type who are cross trained in execution, but think from perspective of the end user and "optimize" towards the best user experience from each media type. You can't compare search to social to display to video using same performance metrics. Problem is, it's expensive and talent pool scarce. But to me, it spells lots of opportunities.

  2. Alejandro

    It's a great article and I agree that last touch attribution is flawed in terms of its ability to "explain" how media drives actions. Ultimately, I think that when we ask for better metrics, what we are really asking for is a better "story" to explain a very complicated process. Depending on how well constructed they are, "stories" that explain very complicated events, like why large numbers of people buy stuff, may be called "theories" or "myths." Positing last touch as the sole driver of sales belongs squarely in the "myth" category. Clearly, a better explanation for why sales happen is necessary; the problem is that any proxy for revenue generated (i.e. relation between impressions served, cookies deployed, and pixels fired) is going to be another "myth" that explains how advertising is working. Maybe a better one, but ultimately, still a mechanism that can be gamed by clever people with money on the line. Until someone much much smarter than I can figure out how to tie revenue directly to ads, I think we'll have to try our best to understand how the process of delivering ads works (as Alex mentions in the comment above), gut-check it (i.e. does serving display inventory purchased at $0.25 CPM at low frequency on sites with 10 ads on a single page really drive revenue?), let this common sense dictate our strategy, and have faith in good creative and good ads.


Add a comment

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>