Bots, the sinners & saints of advertising
Back in March 2017, hot on the heels of the YouTube advertising debacle, Procter & Gamble announced that it was going to slash its digital adspend.
With John Wanamaker’s famous words ringing in its ears, the marketing world held its breath. Was P&G’s decision set to portend a dramatic decline in sales, or was the industry that had so triumphantly toasted digital’s supremacy over its inefficient above-the-line counterpart about to be exposed for selling new clothes fit for an emperor?
Well, news came in late July that P&G had beaten quarterly earnings expectations on the back of a US$140m digital advertising cost saving. That amounted to nearly a percentage point of additional profit margin for the quarter, with reductions to agency and production fees delivering a further boost to profits. All the more galling to digital protagonists was the news that revenue had grown by 2% in category where rivals’ sales were flat.
In the face of such overwhelming evidence, the doomsayers were quick to call digital out.
The last couple of years have been troubling for Facebook and Google, which take roughly 77c in every US dollar spent online. The YouTube malaise, issues with the way Facebook reported on ad clicks and the general opacity with which GAFA (Google, Apple, Facebook and Amazon) operate have contributed to an environment in which many brands feel compelled to put the tech giants at the heart of their marketing strategies, even though they’d rather not.
But, while you can’t argue with the evidence, P&G’s data only tells part of the story.
Its decision to pull spend was based on concerns that its ads were being clicked on by bots. And P&G’s uptick in sales over the past few months suggests this was the case.
Bots create the problem — but offer the solution
So why did this happen and what do we do about it?
For starters, it’s pretty unlikely that GAFA has warehouses full of computers that click on ads flighted on their networks. Attractive as this theory might seem to conspiracy theorists, the tech companies have simply got too much to lose if they disenfranchise their primary advertisers.
The bot clicks were mostly taking place on websites which’d got themselves listed on a variety of programmatic exchanges. As most publishers work on an impression-based revenue model, it’s in their interests to demonstrate that ads flighted on their platforms have been shown and interacted with loads of times. Even if this means cheating.
Any human visiting many of these sites would see that they were pretty crappy and extremely unlikely to deliver the type of sales generating engagement brands such as P&G are after.
Smash the machines
Which is why some companies, such as JP Morgan Chase, have cut the number of platforms they flight ads on from 400 000 — largely picked by machines — to 5 000, all picked by humans. This approach makes sense. But to disregard programmatic advertising in totality, which some have been calling for, is a reactionary perspective that has more in common with the 19th century Luddites than that of an organisation with tech at its core.
Machine learning and artificial intelligence, on which programmatic is based, is in its infancy. It’s inevitable that, like the kid who’s just had the stabilising wheels removed from his bike, it’ll have the odd headlong dive into a thorny thicket.
We all need to make mistakes in order to learn. But, to do so, we also need to be taught. Slating the educators or the curriculum might, as in P&G’s case, appeal to short-term shareholder pressure, but it’s unlikely to shift the needle in the long term. To do that we need to back technology. Yes, we need to stamp out fraud but, instead of blaming the machine, let’s invest in the systems and processes that’ll help it learn.