In our work we’re often challenged to optimise existing marketing campaigns that have a strong digital focus.
And of all the enthusiastic channel managers we meet, across POE (paid, owned, earned media), none are more bouncy than the programmatic video and display advocates.
Programmatic (or prog) is the darling of an industry obsessed with competing for mere seconds of attention from audiences. Their game – working out what it takes to grab and hold the attention of millions of human beings sitting on sofas multi-screening, commuting on trains, ‘snacking’ on social media between work shifts.
And while it’s tempting to throw a million low-cost impressions per day out into the open exchanges, what are we really achieving?
The protest swiftly comes: “just look at the view-through stats on this campaign! Better than generic PPC, better than affiliates…” and only stops when they notice you’re still methodically stirring your tea instead of picking your jaw up off the floor.
And each time, delicately – we have to ask: “what is your market share?” And once we establish that it’s high enough, it’s time to stop talking about causation, and go back to correlation.
As anyone knows, it’s quite easy to correlate ridiculous things like the divorce rate in Maine against per capita consumption of margarine (credit: TylerVigen.com)
The trouble with using programmatic with market share of over about 5% is that you’re simply dropping cookies on people who were probably going to buy from you anyway. And depending on your CPMs and your budget… that could be tens of millions of cookies. You could be sprinting to the marketing director’s office claiming attributable online sales north of half a million units, with a persuasive case to divert more spend your way.
But you could be horribly, horribly wrong.
A better move would be to implement a simple check on programmatic efficacy: the placebo test. Divert 15% of your channel budget in programmatic to PSA ads such as public health or charities, in identical units and using carbon copies of the target audiences your real ads are showing to.
By analysing data on the lift between placebo vs your latest & greatest ad creative, you can start to understand how many sales were down to tagging people who were about to convert anyway, vs those you influenced on their path to purchase.
Another useful cross-check can be found in the viewability statistics for the same campaign. If you’re hovering around the 50% mark, and using the IAB definition of 50% of pixels for one continuous second, it can be quite a stretch to attribute view-through sales to what amounts to a mere flicker on the edge of a human being’s perception.
So focus on lift vs baseline, and walk with grace to the marketing director’s corner office, because now you have the stats to back it up.