“The more things change, the more they stay the same”. So said novelist and journalist Jean-Baptiste Alphonse Karr (albeit in French), highlighting, of course, that as tumultuous as times are, nothing really changes. He probably wasn’t talking about the world of digital advertising and the rise of programmatic buying, but he could have been.
You read ExchangeWire and so you will already be all too aware that display media is in a period of massive upheaval. The move to a new method of trading has resulted in far-reaching changes in many of the ways the industry works. Agencies are moving from providing efficiency through economy-of-scale based media-buying to data-based efficiency. The blind-network ecosystem, once in rude health, is now facing up to reality: adapt or die.
However, beneath all this change, the industry is still faced with the same old challenges, as static as ever. Namely, good old attribution. Anyone who attended 2012’s ATS London event will doubtlessly recall the client panel. For those not there, I’ll summarise: clients still don’t really buy post-view as a metric. It’s understandable, for years pretty much every player in the ad supply chain has been desperate to keep the value of the post-view conversion alive. Whenever a client asks for justification or evidence, there is a tendency to present the same ubiquitous study (thanks comScore!) and hope for the best. Turns out advertisers don’t always believe everything they read. Who knew?
The problem with all of this is that the post-view metric, or some equivalent of it, is vital to properly understand the value of display. If you think it’s a worthless metric you probably think display is a worthless form of advertising. If display is ever going to get a proportionate share of marketing budgets it needs a measure of triggered behaviour change, and users that click just don’t make the numbers stack up. But by blindly telling them that they should accept every conversion tracked within a 14-day period (or 30 or 90 or whatever you convince yourselves you can get away with) the reaction of most advertisers is to instantly distrust the metric and, worse still, the channel.
So what is the solution? A wholesale abandonment of the CPM buying metric and a move away from impression tracking? That is what some within the personalised retargeting business would have you believe. They are one of the most prominent categories of networks to sell on a CPC: only charge the client for clicks, optimise the hell out of the clickthrough rate and forget about the whole post-impression challenge. Only focusing on a CPC model is just ignoring the problem. Even when media is bought on a CPC rate, it can be very hard to justify any significant spend levels purely on a post-click basis. That clickthrough rate may be good, but your effective CPM still usually bakes out to a pretty familiar-looking figure. In other words, the media is costing you the same amount, so why does the model matter? We should all be optimising to maximise engagement, of course, but that doesn’t mean it makes sense to turn your back on every user who doesn’t have the time to click right here, right now.
What is worrying is there seems to be an industry-wide case of apathy. Some networks, distracted by the evaporation of their business as a result of the Agency Trading Desk model, have given up worrying about the debatable value of post-view. “Let them buy CPA!”, they cry! Or instead, they avoid the question by launching a video offering, claiming they are driving brand-metrics when really they are doing little more than demonstrating engagement rates. There is a decreasing appetite for actually demonstrating the value and effectiveness of digital advertising. Whenever the value of a particular metric is called into question, the industry just moves on to a new metric. What we should be doing is actually demonstrating value in the metrics we already have.
The post-view metric has created both appreciation for, and cynicism of, retargeting. Measured one way it looks fantastic, yet viewed through a lens of skepticism, it seems like a chronic waste of money that steals credit from other channels. The Facebook Exchange has been criticised on these very pages for the renewed stimulus it creates for post-impression arbitrage (and rightly so, impressions are cheap and the ad formats virtually invisible), but does that make it okay to write-off an entire channel, or regard 100% of post-view responses to such an ad format null and void?
Perhaps the industry is focusing on the wrong thing. Forget the buying model and stop trying to second guess how consumers respond. Stop looking for an easy fix, a quick study to prove something and get away from difficult questions.
The key to running display properly is to actually have the guts to understand it. Test it out, experiment, adjust and learn. Convince advertisers to invest, in order to understand. Advertisers are totally correct to have concerns — awarding every post-impression conversion (even the de-duped ones) to display is nuts — but it’s no crazier than ignoring every single one of them, and it’s probably less nuts than blindly pumping budget into branded paid search terms without understanding the value there either. Everybody buys things sometimes. Sometimes people click, sometimes they don’t. Second guessing consumer behaviour – assuming they click, or that they don’t – is likely to leave you confused and investing in the wrong place. Let the data tell you how consumers are responding and use this understanding to dictate your marketing investment.
Testing the incremental value of display impressions should be fundamental for any medium- to large-sized display client, just as we should all now be looking at the impact of ad visibility on incrementality. At LBi, we have been able to conclusively demonstrate the impact of display for a number of advertisers, and the knowledge of the incremental value of post-view has led to massive growth in cost-effective investment. However, we can’t assume each result measured is definitive — things change. It’s time the industry started to really understand how customers react to differing advertising strategies and developed more appropriate measurement techniques. Forecasting the value of post-view might be a massive challenge, but it isn’t (as some would have you believe) impossible. Figure out for each different segment how many conversions you would expect from a finite audience with no activity, and use that as the floor from which you measure.
As an industry, we should focus on not taking the easy route or avoiding the question. Let’s try to actually understand something.