Introducing 'vCPM': The Right Way To Think About Viewability

By
  • Facebook
  • Google Plus
  • Twitter
  • LinkedIn

adrian-t-dataxu

“Data Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media. 

Today’s column is written by Adrian Tompsett, VP of Business Development at DataXu.

Viewability is on the top of every digital marketer's wish list for 2013. The emerging metric seeks to measure how frequently an ad is actually seen by a user, and not merely the number of times it’s served by a site’s ad server. It’s safe to say that all people (except maybe our cookie bombing friends) agree that viewability is a good thing for the industry and a step toward pulling real TV dollars into our space. Ad networks in particular built their optimization strategies around making sure their cookie was last in line to receive view through credit. Bringing viewability into the mix offers the opportunity to legitimize this often (and rightly) criticized attribution model.

Recently, a spate of articles have examined the various viewability measurement methodologies, outlining the pros and cons of each in order to help marketers determine how to factor viewability into their digital media planning for 2013.

Unfortunately, these articles – and the industry dialogue as a whole – miss a significant point: viewability is a metric that must be evaluated relative to cost in an eCPM, and not just as a percentage of impressions in view. The time has come to add another acronym to our vocabulary: vCPM.

What marketers can learn by vCPM vs. number or percent in-view

Let’s walk through a sample scenario. Say you have $100K to invest in one of three digital campaigns, and your goal is to purchase viewable impressions for your brand. Your options all cost exactly $100K, and they’re behind three doors:

Door #1: 25MM confirmed views

Door #2: 20MM confirmed views

Door #3: 9.3MM confirmed views

Which campaign option would you choose? If you’re measuring the number of confirmed views, Door #1 seems like a no-brainer, right?

Now let’s say you plan on making your selection based on the percentage of in-view impressions, and that these campaigns offer the following scores:

Door #1: 25% in-view

Door #2: 50% in-view

Door #3: 75% in-view

Still want to proceed with Door #1? A 25% in-view score feels rather abysmal, and I doubt any marketer would happily sign up for a campaign in which 75% of the impressions are served out of view.

But here’s the kicker: percentage of impressions in view doesn’t tell the whole story. Let’s add more campaign details to see how this all comes together. As in real life, each door charges a different CPM, which drastically affects the number of impressions you can buy for your $100K, and that in turn affects the percent of impressions in view:

dataxu-viewability-chart1

This table should give any marketer pause. Simply selecting an inventory source based on the percentage of in-view impressions can lead you down the wrong path, causing you to spend your way to an inferior ROI. The upshot? Rather than worry about how many impressions are ‘wasted’ by being out of view, focus on how much you pay for impressions that are confirmed in-view. Door #1 is looking pretty good once again!

Regardless of the methodology your publisher or partner uses to measure its viewability, you can still calculate vCPM using this formula:

Total Spend / ((Total impressions * % in-view) / 1000) = vCPM

$100K / ((100,000,000*.25) / 1000) = $4.00

Going forward, marketers need to hold publishers and partners with whom they work, and the tactics they use, accountable to both viewability performance as well as performance relative to cost. The vCPM is the best way to get a truly apples-to-apples comparison of viewability effectiveness.

VCPM is hardly ‘spray and pray’

Now, if you’re wondering whether the vCPM approach is nothing more than "spray and pay" – where marketers go to the ad exchanges in an effort to buy as many cheap impressions as they can and hope for the best – the answer is unequivocally no.

Writing off vCPM as spray and pay implies that higher-priced inventory performs better than cheap inventory as a rule. But is that the case? It is for certain activities, such as credit card signups or consumer electronics purchases. After all, those consumers who are identified as more likely to complete those actions become attractive prospects to more marketers. Over time, competition for the impressions served to those consumers heats up, leading to an escalation in price. That’s why In-Market or Remarketing tactics earn comparatively high CPMs.

But, we don’t have a third-party data segment called ‘In-Market for Viewable Impressions.’ Viewability is  a performance activity that can be found across the entire digital sphere, in impressions delivered to all kinds of consumers visiting all kinds of sites. As a result, “viewable” impressions aren’t inherently more expensive, just as expensive inventory doesn’t have an inherent vCPM advantage (as will be discussed below).

What’s more, because the measurement is still evolving, deciding at bid time whether an impression will be in-view remains an inexact science. In this light, one could even argue that the cheap spray and pray approach offers a better path to vCPM success than its more expensive alternative, "pay and pray."

Why expensive inventory doesn’t always have a vCPM advantage

The viewability metric is different from other performance activities in another key way: those activities are limitless, whereas viewability is capped at 100%. To use a sports comparison, think of bowling: no matter how skilled your opponent, he or she can only score a 300. Likewise, the most expensive, premium inventory on the web can only achieve a maximum 100% in-view.

Using our campaign option examples above, let’s pretend that Door #3 was actually 100% in-view and therefore delivered an $8.00 vCPM. Even with perfect in-view performance, the burden of Door #3’s significantly higher cost allows both Door #1 ($4.00 vCPM) and Door #2 ($5.00 vCPM) to be significantly more cost effective. More simply put: the marketer who bought Door #3 is paying more for each viewed impression, and getting fewer of them.

Viewability performance, like all performance metrics, should be evaluated relative to cost

Without including cost in your performance equation, there is no way to fully understand and optimize the viewability of your media plan. Furthermore, smart marketers will recognize that this metric is but one of many metrics that matter. The task of optimizing advertising ROI is a multivariate problem, and the modern marketer needs an approach and supporting tools that enable them to not only isolate and test variables, but also to pull it all back together in an automated way that weighs the various sensitivities and trade-offs for the best result for their brand. Marketers who get this, and jump up the learning curve quickly to solve for multivariate advertising ROI will be the biggest winners in 2013, and the years to come..

  • Facebook
  • Google Plus
  • Twitter
  • LinkedIn

Email This Post Email This Post

By on at

12 Responses to “Introducing 'vCPM': The Right Way To Think About Viewability”


  1. Alejandro says:

    A "viewable-CPM" metric is definitely a step in the right direction, and hopefully we'll be able to tie view's to conversions (and conversion funnels) in the near future. One issue your article didn't address is the reliability of the technology that measures viewability. Not too long ago adexchanger reported the MRC was not fully satisfied with the way viewability worked:

    http://www.adexchanger.com/analytics/iab-and-mrc-no-viewable-impressions-are-not-ready-for-prime-time/

    I understand that there are many stakeholders in this conversation, and that each has different agendas. With that said, I would be wary of adopting a technology like this until I was fairly certain that it worked, particularly across exchanges where impressions get bounced around so much.

  2. Barry Scott says:

    Isnt it still impossible to track viewability in a scalable way across adexchanges because of adexchange tags being implemented in iframes? Adsense tags are all in iframes per standard I believe, so unless theres a way to measure viewability of iframes I dont think theres a scalable solution? Correct me if I'm wrong, but all the solutions proposed so far involve custom changes on the publisher side, which wont happen in a quick and scalable fashion (no incentives for pubishers to make these changed).

  3. Tom Shields says:

    I agree that vCPM is a useful metric for brand marketers, and you provide some insightful examples, Adrian. However, I think you are wrong that "It’s safe to say that all people (except maybe our cookie bombing friends) agree that viewability is a good thing for the industry" - I was on a panel last week with decidedly mixed opinions, and not just from cookie-bombers.

    Also, I don't think "deciding at bid time whether an impression will be in-view remains an inexact science" - it's actually impossible to know. By definition, you can't measure viewability at bid time, you have to wait a second and see if the user scrolls or clicks away. So vCPM belongs in the same post-delivery metric bucket as CPE or CPA, and while useful metrics like they are, should not be seen as replacing CPM as the primary currency of display advertising.

  4. Thanks for your comments, Tom. And agreed - given the definition of viewability, it is impossible to know before you bid on an impression whether or not it will be viewed. This is an important point to make as marketers get familiar with a new metric.

    More generally, from my perspective, viewability is the newest parameter (of perhaps a never ending string) to come to market this year. On the one hand, I agree with you, this is nothing more than another cost per fill in the blank here.

    On the other hand, the big 'a hah' is about cobbling all of these different (and sometimes competing!) metrics together to pop up a level and take more of a universal & holistic view. One that enables not only the analytics to help marketers understand this new world, but also to provide the 'fuel' for multi-variate optimization in order to deliver real marketing effectiveness.

  5. Barry, you are correct in that viewability measurement across the exchanges remains a complex problem. Article referenced by Alejandro is a great place to start.

    http://www.adexchanger.com/analytics/iab-and-mrc-no-viewable-impressions-are-not-ready-for-prime-time/

    That said, end of the day, digital media is always measured view 'sampling', whether you are measuring your CPA or validating the audience your impressions were served to on a given campaign. Imperfect measurement is almost always better than no measurement, provided you understand the underlying methodology and can interpret the results given known limitations.

  6. Kirby Winfield says:

    @Adrian - Great piece. It's nice to see the focus placed on the immense benefits of viewability measurement as opposed to the few flaws. The good outweighs the bad heavily.

    @Tom - Actually there is no reason, save the lack of technical resources committed by ad server owners (proprietary, third party, et al), why viewability cannot be actionable in-delivery. At AdXpose we built an ad server that decided to deliver the ad based on whether or not the placement had been scrolled into view.

    As for RTB - bidding engines could be tweaked to allow for vCPM "hold" bids, which would override lower non-vCPM bids in the event an ad became viewable, while simply not triggering (and serving the lower priced ad) should an ad not become viewable.

    Of course, that would inherently require RTB entities to transparently show the percentage of inventory that is never viewable - or put another way, to admit they knowingly serve ads that are never viewable - so let's not hold our collective breath.

    • Barry Scott says:

      "Of course, that would inherently require RTB entities to transparently show the percentage of inventory that is never viewable"

      true, but that also assumes that said RTB entities are charging their advertisers on CPM or on a post-view basis right (e.g. post-view CPA)?

      RTB entities/DSPs that charge on CPC or CPA targets, post-click, should benefit in the long run here if viewability becomes a huge issue for advertisers. In these cases viewability is more an issue that DSPs should raise with Ad Exchanges to solve as they are buying off them on a CPM basis.

  7. Matt Sherman says:

    Always look at the actual delivery numbers and/or proposed delivery numbers. The percentages don't matter. Same goes for publisher revenue share - 70% of $1CPM is less than 50% of $2CPM.

    It's very surprising that this still needs to be explained. Nice explanation Adrian.

  8. Jay Wright says:

    In the end, a buyer will evaluate the "effectiveness" of the campaign based on how much they paid vs. how many valuable outcomes (however they define those) it delivered.

    So what if "Door 3" in your example delivered 3x the value of the other campaigns? Wouldn't that make it the most cost effective?

    It all comes down to how you measure your "R" in ROI. You can really put anything you want in the denominator, but a smart buyer will realize that the denominator has to be the thing you are trying to maximize. For a branding campaign, you are probably trying to maximize a delta in some "brand awareness" metric, and if you have a DR campaign you are probably maximizing clicks and/or post click actions.

    In your examples, you have defined "view ability" as your "R". I just question how that fits into a marketers goals. Intuitively, this debate can be framed as a step forward, but to me it seems backwards. It feels like vCPM fans are proposing to regress into measuring a proxy when the real thing is already available. I think most marketers get this and that explains the sluggish and spotty adoption of the metric.

  9. Alejandro raises an important point about post in-view response. Viewthrough.org seeks to address this by pushing for standardization of this metric's definition, scorecarding vendors and methodology best practices.

    Thinking about ad viewability and see-through without also thinking through viewthrough leaves the door open for clickthrough rate inflation.

Leave a Reply