Why Viewability KPIs Are Bad for Programmatic Performance
In his opening talk at AdExchanger’s Industry Preview last week Gary Vaynerchuk brought up the conflict that often exists between metrics and results: “The problem is that the metrics and the sales don’t match up,” he said. “That’s got to be talked about…”
One key area I see that happen a lot is with viewability rates. Let me define what that means first: A viewability rate is the % of time that a particular ad unit on a particular URL is measured as In View after the ad is rendered. It was hard listening to the Ad Verification companies at Industry Preview talking about transacting on “viewable” inventory. I’m not a fan of that term. Aside from blatant fraud any ad on any website certainly can be “viewable”. There is no way to know with 100% certainty that any ad will definitely meet the MRC definition of a viewable ad before it is actually served.
A more appropriate measurement what is the average rate at which a particular ad unit on a particular website is measured as “in view”, aka the viewability rate. With that, you can certainly predict an average viewability rate and use that to guide buying decisions, but any user can X out of any page at any second so there is not “viewable” verse “not viewable” inventory other than outright fraud. Even an ad placement that is in view 99% of the time will sometimes not meet the MRC standard. So for the purposes of this article, I am focused on viewability rates and how using that as a metric can hurt campaigns with performance objectives.
It has been evident over the last 2 years that the average viewability rates of ad opportunities are driving pricing in the programmatic market. We have seen that the available inventory across all major exchanges is roughly inverse to the viewability rate being targeted. For example, to achieve a 70% viewability rate a campaign is limited to approximately 30% of RTB opportunities. The result is that ad placements with higher average viewability rates trade at higher CPM rates – anywhere from 50% to 3X. Ad placements with low average viewability rates, on the other hand, are increasingly cost-effective opportunities often well worth the effective CPM for only those ads that end up being viewed.
Striving for viewability is a KPI worth considering for branding campaigns because it is obviously critical to having any hope of impacting brand recall. Also, if an advertiser is buying site-specific placement where the only qualification is reaching a visitor to a particular website then viewability also makes sense as long as you can afford the premium CPMs.
While there are many cases where viewability is a worthwhile KPI when it comes to running programmatic advertising for performance-focused campaigns that target specific users it is counter-productive and raises CPAs. We have seen this time and time again on our platform across channels and advertisers.
So how can using viewability as a KPI hurt performance?
Viewability requirements on a programmatic ad campaign that is optimizing based on user attributes (browsing behavior, purchase history, demographic data) hurt performance by eliminating ad opportunities against highly desirable users that are most likely to drive results, just because of the probable viewability rate of a particular ad placement. Additionally, the premium CPMs required for media acquisition due to scarcity of highly viewable inventory against desirable users inevitably increases the CPA for advertisers without getting a proportional lift to their bottom line.
This article from the adexchanger covers a study by the Goodway group on viewability and the economics of it. This is another article specifically about the downside of 100% viewability. On the ownerIQ platform campaign results have definitely shown this statement from the Goodway Group Research to be true for most of our advertisers:
“Although it [Goodway Group Research] found that viewability improves conversion rate by 8% to 9%, that tapers off fairly quickly. Sites with between 30% and 40% viewability rates were the best value for money in terms of CPM paid to conversion generated. For sites with 80%-plus viewability, the conversion did not justify the CPM.”
There is no way to know for sure before an ad is actually delivered whether it will be measured or not, and if it is, whether it will end up being counted as In View, aka viewable. That said, we have found that the average viewability rates of particular ad placements are generally consistent. So while you can’t know if any individual ad will be measured, and if it is whether it will be In View, over a significant number of impressions the viewability rate is predictable.
I have been on a couple of client conversations discussing the trade-offs on a conversion-goaled campaign that prioritize viewability, particularly under our targeting models. On one call the client actually asked this question: “How is it possible that ownerIQ can deliver the best CPA on our campaign, but no one sees your ads?”
While that was a bit of an exaggeration by the agency, their point was based on the invalid idea that in all advertising viewability drives conversion rates. If you are just buying a publisher-specific buy with no audience targeting then viewability makes a difference. Not so for programmatic. Here’s how I answered that question from the client and got them to see the light:
“Imagine that we have 100K opportunities to deliver ads to users with very high EP scores (meaning they are a great match for the targeting profile we built on your existing customers), but the average viewability rate of those opportunities is 50%. Now, because our targeting algorithms make decisions on proprietary brand and product data most of these prospective customers were uniquely selected by ownerIQ, so delivering ads to these users is even more important because our competitors will not be targeting them. Without ownerIQ reaching these users the conversions will be lost, as will the customers, and the lifetime value they bring to the table.”
With an arbitrary goal of 70% viewability, we would not be able to deliver ads on opportunities with a 50% viewability. As a result, none of these high-value, unique customer prospects would see ads for the client. However, without the arbitrary requirement for viewability, we would reach 50K of the prospective customers. Even at a conversion rate of 1%, the client will generate 500 new customers. With the viewability rate requirement in place none of these customer convert.
So what does the client want – 500 incremental customers or a 70% viewability rate?
In the end, the client wants results. Driving to get more ads seen is a good objective, but it should not trump getting results by passing on opportunities to put ads in front of users just because they have a lower chance of being viewed. This may not be one of the metrics Gary had in mind during his talk, but it certainly fits the bill.