Attention Economics Pricing In Advertising: Cost Per Second

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 Matt Shanahan, SVP of Strategy for Scout Analytics.

Engagement is the unit of monetization in the ad world, yet engagement is not what is bought and sold in online advertising. Instead ad placements are priced and sold based on impressions. Whether that impression lasts 1 second, 1 minute, or 10 minutes, the price of the impression is the same which doesn't make sense in attention economics.

The reality is an audience member is not likely to consider an advertisement until the publisher has truly caught his or her attention, i.e., until the person is engaged. Consider an interstitial advertisement that comes up between a search result click and viewing the content. The audience member is interested first and foremost in figuring out if the content is even relevant. There is little or no chance that person is willing to sit through an ad when they don't even know if the content is relevant. If the audience member does make it to the page, the person will quickly scan the content to determine relevancy. If it is irrelevant, guess what, the person moves on without considering the ads that were served. If, on the other hand, the content is relevant, the person will spend time on the page and often considers contextual content provided, e.g., relevant ads and referenced content. Think about it in other forms of media whether TV, print, radio, or live events, the producer gets the audience engaged and then inserts the advertising.

In many ways, an online publisher is penalized for providing a highly engaged audience member to an advertiser. If the publisher creates an engaging article that is actually read, the publisher gets revenue from an impression but maybe not from a second impression. This gives rise to publishers that game the system by chopping up the content into multiple clicks to create more impressions but that hurts engagement.

Stop the madness. Stop selling impressions. Advertisers, publishers, and audience members alike are better off buying and selling advertisements based on seconds of engagement. For an advertiser, buying seconds of engagement means the quality of the impression is greatly increased and the attention getting opportunity has better probability. Selling engagement means a publisher can be rewarded for the length of time they create for the advertiser (i.e., merit-based revenue). For the audience members, they get a better more engaging experience rather than SEO optimized, click-gamed sites.

Engagement is measurable and predictable. Every audience member has a pattern of behavior regarding frequency, context, and visit length. An audience member can be tracked and predicted as to how frequently they return daily, weekly, monthly, etc. The audience member can be tracked and predicted as to why they return such as topic, breaking news, author, etc. And of course, the audience length of visit can be tracked and predicted.

Because it engagement is measurable and predictable, it can be inventoried and sold. Cost per second (CPS = CPM*1000/expected length of engagement) becomes the price of a delivered advertisement. CPS rewards publishers for creating engagement rather than impressions.

Why is CPS important to publishers? Revenue optimization. At any given point, the revenue capacity of a publisher is limited by the audience size. Yes, the audience can be built overtime but that is overtime. No the audience size cannot be infinite so there is a limit. Like it or not the vast majority of audiences are niche. In a revenue capacity constrained model, maximizing revenue per unit is the key to success. Impressions reduce the number of units a publisher has to sell. Engagement increase the number of units. Given engagement is the unit of monetization, engagement is what publishers need to sell. Engagement is finer grained than impressions and therefore can provide revenue optimization opportunities.

Why is CPS important to advertisers? More control. Using CPS, advertisers can control the quality of the purchased ad placement. Right now, insertion orders are governed by CPM and minimum impressions and maximum spend. In the CPS world, an advertiser would specify, CPS, minimum impressions, maximum impressions, and maximum spend. Essentially, giving advertisers control on length of time for their impression.

Why is CPS important to audience members? Return on time. Quality experience and content relevance are the ingredients that make an engaged audience member. CPS incents publishers to make the investments that benefit audience members.

CPM is the wrong pricing in attention economics. It promotes revenue activities that are not beneficial to advertisers or audience members. CPS is the right pricing in attention economics. CPS aligns the motivations of publishers, advertisers, and audience members alike.

Follow Scout Analytics (@ScoutAnalytics) and (@adexchanger) on Twitter.


  1. Andrew Boer

    I don't understand how a CPS model actually reflects engagement with an ad outside of video, interstitials, and expanding ads. It certainly wouldn't reflect engagement with a flat display unit or a site takeover.

    If I am engrossed in content on a page with two ads above the fold, I might have a high CPS for the ads without ever engaging with them. The extreme case would be online gaming, where I might spend hours on a single page without engaging with the ads. Widespread adoption of this kind of metric would mean every publisher would gravitate towards content that wastes the most of the users time.

    I think the way SAY Media (Videoegg - who recently acquired SixApart, my employer) approaches the problem is a good middle ground..they base engagement on the mouse over--which then expands an ad. If the mouse over engagement approach became standard, arguably you might eventually extend that further into a cps model -- but for now it seems like an enough of an improvement over the CPM.

  2. While the criticism on current metrics is valid I don't believe time spent is a perfect solution. No matter how engaging the content is, the advertising that accompanies it has to be relevant and appropriately targeted to justify a premium.

    However, if the content itself is branded, say an advertorial, a branded app or a game, then time spent is a valid measure.

    We also need to demonstrate that the engagement translates into some action on the part of the consumer.

  3. Matt,

    Good thoughts, but the data proves otherwise. On a sample size of 1000 premium publisher sites and 1.6B impressions we have actually seen no statistical correlation between time spent and engagement within the ad unit:

    *Metric* *Coefficients*
    Intercept 0.006269938
    AverageMinutesperPage 0.005794597
    AverageMinutesperVisit -0.00010584
    AverageVisitsperVisitor -0.00039917

    We, like Andrew and VideoEgg, value time spent and engagement *within the ad unit* as true media efficiency metrics.

    @Milind - see here for engagement driving action by consumer:

  4. This is silly.

    For one, the idea of cost-per-second becomes less valid as more users adopt the browsing behavior of multiple tabs and leaving them open for long periods.

    And to echo @Milind's point, this proposal would be like billboard sellers counting the cars that drive by and charging more for people stuck in traffic. Sure - exposure probably goes up - but the real question is, did those consumers go on to buy the item/service advertised? That is something that actually can be measured online and should be.

  5. The points above are good to consider.

    To Andrew's concern about integration of ad and media, it is up to the publisher to architect the best possible opportunity for engagement through layout, sequencing, format, etc. This doesn't take away from the value of measurng and pricing by length of impression.

    To Kirby's concern about granularity of impression, it is up to the publisher and advertiser to decide if impression length will be measured at page or ad level.

    To Josh's concern about multiple tabs, tracking codes can figure out if a page is visible or not -- that has to be factored in. Additionally, you bring up a statistical edge case that most consumers do not do.


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>