David Morris of CBS, the incoming IAB chairman, established those two big issues right out of the gate, in his opening keynote: (1) the viewable impression; and, (2) fraud and the trustworthy supply chain. He went on to say that "Publishers need to demand that verification companies do a better job of delivering consistent and accurate data." He noted further: "The viewable impression is a new measurement standard that we are trying to get right and get accurate….. [b]ot traffic is illegal activity and those who knowingly use fraudulent traffic to take advertising dollars should pay a price."
If I can contribute one small perspective to the industry dialogue, it is this: These are NOT separate issues. Remember the lesson from the film “All the President's Men”: "Follow the money”? The money in fraud is in impression generation. Fortunately, the MRC has made it clear in no uncertain terms that a fraudulent impression is, a priori, not a viewable impression.
There has been a lot of scrutiny of how different viewability providers treat things like cross-domain iFrames, WebKit browsers, and so on. How much scrutiny has there been on these same providers' ability to identify and exclude from the viewable impressions tally, fraudulent impressions? No one wants to pay for a viewable impression logged by a fraudster, yet that is precisely what is happening today.
Like it or not, fraud and viewability converge, because successful execution of fraud results in the generation of faux viewable impressions. And the viewability providers that will survive the inevitable thinning of the herd will be the ones that can filter the fraud.
Indeed, Morris called for a thinning of the herd.
On Monday afternoon there were four concurrent industry town halls; of course I went to the viewability session. Emotions ran hot from both the sell side and the buy side. Not surprisingly, the agency representatives balked at the IAB’s proposed 70% threshold for the share of paid impressions on a campaign that should be viewable. To be clear, no publisher can guarantee that any given impression will qualify as viewable -- those pesky users on the other side of the screen still have the power to scroll or click off before the ad logs a second of screen time -- and no publisher can therefore expect 100% of served impressions to qualify as viewable. But the 70% threshold refers to the share of paid impressions, not served impressions; the agency position is, if I buy 100 impressions and 85 are viewable, I’ve got another 15 coming.
Of course any time you get buyers and sellers in a room together, they are going to end up negotiating. The agency people at the session seemed willing to work with publishers to find common ground, where common ground may be found. Mitch Weinstein of Magna said his company would accept viewability data from whichever vendor the publisher is using -- “as long as they’re MRC-accredited, and we’ve heard of them.”
There was a general consensus in the room that viewability -- 30% or 50% of pixels on the screen for 1 or 2 consecutive seconds -- was just the beginning. Once we’ve successfully integrated the viewable impression across platforms into the everyday work flow, only then will we be ready to begin focusing on “metrics that matter” (a tip of the hat to IAB’s Sherrill Mane), like effectiveness and engagement. Getting the ad on the screen isn’t the end of this journey; it’s the beginning.
To be sure, there’s some good news on the viewability front. Publishers are working to optimize their inventory against viewability, and vendors are providing tools to support that effort. And while we constantly read the headlines about half the Internet’s inventory or more being not viewable, the fact remains that for premium publishers (and here I invite you to propose your own definition of what “premium publisher” means -- over Scotch Sunday night we could not agree), the viewability picture is pretty good, and getting better.
Of course publishers don’t even want to hear that 5% of inventory isn’t viewable if that means they can’t monetize that 5%. But the premise behind viewability all along has been that the bad actors will be exposed and driven out, and that demand for high-performing inventory will more than compensate for the leakage from non-viewable impressions.
Here's a thought.
The problem is not viewability. Rather, viewability is a symptom of the disease which is CPM.
For more than 20 years our industry has been complicit in a massive fraud - the meaningless use of CPM as a metric that really does not matter. Applying a print model to online was never going to work. If you take CPM out of the equation - the need for viewability becomes redundant.
5 years ago the company I ran tested alternative models and to some accounts defined the term 'metrics that matter'. We ran display media campaigns priced according a simply, yet, meaningful, set of metrics. The first was a cost per engagement. Publishers only charged advertisers for an ad that received engagement - in this case a 1 second or longer mouse over or a click. Following the initial point of engagement, Publishers charged advertisers for the time users spent interacting with the ad. This of course necessitated the creation of ad units designed to keep users in the ad, on the publishers site, thereby removing the need for click throughs.
It always stuck me as nonsensical that Publishers build capital through generating traffic and then monetise that traffic by trying too get users to leave the site.
The ad units we built were self contained contain channels that delivered on average more than 2 mins interaction time. In some instance, such as live data feeds from Sporting events, it was not uncommon to receive interaction times of 30 mins or more.
The third metric we applied was based on end actions - these being defined by known business outcomes specific to the advertiser. For example, in Auto, car configurations, brochure downloads and dealer requests were often sited - all of which were completed inside the ad unit - for which the advertiser remunerated the publisher.
The fact that the industry is still debating the definition of a completely meaningless metrics suggest to me that as an industry, we are all complicit, in a massive fraud. As David Morris said. Follow the money...."
Unfortunately, branding advertisers do not operate by direct response rules. They need to get their message across to all targeted consumers even if said consumers do not take immediate action that can be measured electronically. Indeed, the overwhelming percentage of sales responses to branding ad campaigns occur due to their cumulative effects, not because of a single ad exposure. If a digital ad seller is going to charge branding advertisers only when the response can be monitored via a click or some other " engaging" action, but not guarantee that the rest of the ad "impressions" are real---and these are, most likely, 95%+ of the total-----then forget about branding ads entirely and stick to direct response. Just my opinion.
Agreed with Ed - you can't rely on direct (e.g. click, hover) metrics, even for DR campaigns. Nielsen showed little to no correlation between those metrics and sales... in 2011. They did, however, show a positive correlation between online advertising and brand health metrics (awareness, consideration, etc.)
Fractional attribution solutions are the key to unlocking performance, but there still needs to be a common buying metric. CPMs don't guarantee a view (no media channel does) but should get close to guaranteeing the *opportunity* to view.
Some publishers like the Financial Times have been experimenting with new metrics, but for better or worse CPMs still rule the roost.