Skip to content

Viewability, View Through Conversions & Inventory Quality

Share Button

As someone with a background in search marketing, I have always questioned the validity of view through conversions.  While it can’t be denied that passive advertising works (think TV & Billboards), quantifying it in the online space is difficult.

As I work with an ever increasing number of clients from an integrated strategy perspective, correct attribution becomes increasingly important.  A number of years ago, a search agency would have reported on search driven conversions, the display agency, display conversions, email specialists the conversions they assisted in an so forth.  The issue here is that there would often be multiple parties taking credit for driving a single positive outcome, and while they all played a part the sum of the performance of each partner would be significantly greater than the total outcome.  Here enters attribution…

Display agencies, in a typical environment will take credit for all outcomes that arose from after a view or a click on an they placed, with a long look back window.  Now, I agree if you clicked on a display ad and later took action with that advertiser, there is a strong likelihood that your click influenced that decision.  However if you were simply served a display ad, the likelihood that you took an action based upon that ad is significantly lower, however under the default credit model employed by most programmatic & display buying service providers that view was just as important as a click and will be for 90 days.  **Queue Gasps**

But lets take it a step further, what about those ads that are at the bottom of pages that may not ever be seen by the visitor.  Well, by default, those ‘view through’ conversions are counted as well.  There are plenty of reputable sources that place ads at or near the bottom of their pages (think News sites) and its a legitimate placement, as long as its cost is reflective of its positioning.

Taking this concept a step further, there are a number of unscrupulous webmasters who will place ads in hidden locations on the page, either behind content or in hidden elements that are totally invisible to visitors.  Once again, these impressions may also go towards ‘view through’ conversions.

This is the precise reason for the viewability metric which is increasingly being reported upon by both agencies, publishers and programmatic tools.  The definition of viewability differs slightly depending on who you speak with, however the standard definition offered by IAB is:

The industry standard as developed under the leadership of the Media Rating Council (MRC) calls for desktop display ads to be considered viewable if 50% of their pixels are in view for a minimum of one second and for desktop video that standard is 50% for 2 seconds.

This image helps understand the difference between viewability and impressions:

ad-viewabilty-vs-impressions

Viewability should be a fairly good indicator of how likely someone was to interact with  your ad and take action as a consequence.  It should be noted that each individual impression has a binary outcome for viewability (ie. was it ‘viewable’ or not) but campaigns, ads and  strategies has an aggregate viewability score expressed as a percentage (ie. 30% of ads were viewable).

For me, the most basic step in building an attribution model is excluding non viewable impressions.  If an ad wasn’t able to be seen, then it shouldn’t be accountable for generating a conversion.

When it comes to viewability, it would be unreasonable to expect that any campaign experiences 100% viewability.  Ad Blockers, lack of javascript or browser functionality, device type, screen sizes and positioning are all going to limit the viewability of your creative, however, aiming for higher viewability will always be a better outcome for your campaign.  That is, the greater the percentage of your ads that are seen, the higher the chance that they drive consumer behaviour.

A really quick optimisation approach that I often employ is looking at the CPVM (cost per viewable 1000 impressions) at a site level.  Many advertisers are shocked when I first review their data at just how much money is being wasted on impressions that are not viewable.  The formula to use is:

Aggregate Cost / ((Impessions * Viewable %)/1000)

Consider two sites that you’re advertising on.  Site A has a CPM (cost per 1000 impressions) of $3 with a 42% viewability.  Site B has a low CPM of $0.9 but only 15% viewability.  Both sites serve 1 million impressions.

In the above case, Site A has a CPVM of $7.14 & Site B has a CPVM of $6.   While Site A has likely a high quality of inventory on a CPM basis it is significantly more expensive, once you take into account the CPVM, the price of inventory on both sites is much closer and Site B may be a more cost effective buy.

In recent analysis, it’s not been uncommon to find sites who’s visibility is very near single digits that still charge near premium level CPMs, thus having CPVMs in the hundreds of dollars.  Keep in mind that you need a sample of more that a couple of impressions to make any kind of meaningful determination about that quality of inventory using viewability.

A good cross reference metric for site quality is CTR (Click through Rate).  While I’d never advocate click through rate as key metric for determining performance of a display campaign, when paired with viewability it can be used to weed out low quality sources of media.

When advising a client recently we discovered a programmatic partner they’d been working with who was charging a flat $3.5 CPM for inventory.  The deal being that they hit a target CPA and run an arbitrage model attempting to buy inventory for less than they sell it to the client for.  The partner was hitting the CPA target which was a good start, however in their reporting, they failed to break conversions down to view through vs click through, share attribution window lengths, sources of media (domains) and viewability scores.  The did however share the clicks figure, and using that along with the total number of impressions served it was clear the quality of inventory was pretty low.  With a CTR hovering just below 0.01% they were clearing driving the vast majority of acquisitions via the view through mechanism and serving the majority of ads to people who were not really interested or were not able to see the ads at all.

When running a programmatic campaign focusing on direct response, CTRs of 0.15-0.2% should be achievable with fresh creative.

Back to my clients programmatic partner, our assumption was that they were buying bottom feeding inventory for $0.20 -$0.40 and selling it to our client for a 10x markup while relying on a large enough footprint against the Australian population to score the view throughs required to meet the CPA target.

The moral of the story is, if you’re trying to drive direct response but your partner doesn’t share viewability or view through vs click through conversion breakdowns, there may be something fishy going on.

Published inProgrammatic

Be First to Comment

    Leave a Reply

    Your email address will not be published. Required fields are marked *