The programmatic honeymoon is well and truly over. If it isn’t bot traffic, it's viewability, murky trading structures, or just a general disappointment the reality’s not quite matching the promise.
In an attempt to improve display advertising for their clients, Group M announced they’d adopt a 100% viewability stance when it comes to buying impressions.
Unilever said a similar thing. Although laudable in its intent, I believe it's approaching the matter in the wrong way.
Publisher display revenue models are based around having a certain amount of inventory to sell. When you say to a publisher that you’re only going to buy viewable inventory, it writes off a sizeable chunk of what they have to sell. Basic economics says when you reduce supply but demand remains unchanged, the price goes up.
Why campaign viewability doesn’t matter in the way you think
From a trading point of view, this calls into question how pricing will change when buying 100% viewable impressions. However, I believe 100% viewability is the wrong approach. In fact, the viewability of a campaign doesn't matter. Well, it does, but not in that way.
Group M’s aspiration is, understandably, fairly common on the marketer’s side of the fence too. Many want to invoke a 50-70% viewability minimum to all their display activity – some go so far as to have it written into their trading desk contract.
Once again, it overlooks a crucial fact: increasing your viewability increases your CPM, so all your historical "cost per whatever" targets are suddenly going to be harder to hit. It also means that your current targets are most likely based on historical activity, whose results stemmed from a non-viewable impression. So these need to be looked at again and adjusted.
Using viewability percentages to qualify the impressions you are buying is short-sighted. Viewability should be applied to the end point: the conversion. If you only track and apply Cost-Per-Acquisition (CPA) to conversions that came from a viewable impression, then the only factors that become important are viewable CPA and the viewable CPM. When you do this, the viewable impression % becomes redundant.
Let me explain: £100 worth of impressions at a £5 CPM and a viewability rate of 50% (i.e. 10,000 viewable impressions), gives you a cost per viewable impression (vCPM) of £10.00. The same budget spent on lower-cost inventory (£2 CPM) with a lower viewability rate (30%) gives you a vCPM of £6.67.
Assuming you have the same conversion rate for both the above scenarios, you’re going to have more conversions for your budget with the lower viewability rate inventory than you are buying more viewable impressions. This is still true even with a lower conversion rate for the lower-cost inventory - a more likely situation.
So, here it’s actually more efficient to have the positive wastage of a lower viewability rate. Demanding a minimum viewability of 50% would actually be detrimental to the campaign. It works for less performance-focused campaigns too – this approach maximises the opportunity for your ad to be seen to impact any brand metrics.
The 3 key things
So how can advertisers take advantage of this?
1. Calculate your target based on actions that only derived from viewable conversions. Look at your historical data and disregard anything that came from non-viewed impressions. This is your new benchmark.
2. Work with a provider that has the ability to tie a viewable impression to your intended KPI – so that optimisation can actively be done on this basis.
3. Stop setting viewability targets.
To go back to my original metaphor, the marriage between programmatic and display has a great future, but like all relationships it needs work.
Unfortunately, a viewability-focused KPI doesn’t mark the end of the work. Once you have one in place, it sets you up to actively optimise to measuring incremental conversions. This is about identifying the conversions that only stemmed from non-viewable impressions as a basis to understand the baseline activity that would have happened without your advertising. Anything over this baseline shows the uplift directly delivered by your campaign.
But incrementality is a whole other story.