If you’re interested in technology, economics, and human behaviour, you’ll enjoy the podcast conversation between Alex O’Connor and Rory Sutherland (both of whose output I recommend getting to know, if you don’t already).
Note the excellent points made in this particularly enlightening two-minute snippet from 13:56.
Here’s a condensed version of the exchange:
Sutherland: My argument is that we have a weird tech fetish around the optimisation of efficiency, which essentially means eliminating humanity from every process. The assumption in the tech world is that you always eliminate the human and the wage cost from everything.
There’s also this massive assumption in economics that disintermediation is inherently good.
Let me give you an example. I’ve been a campaigner for maintaining manned ticket offices at railway stations. My argument is that there are some functions I want performed by a machine: for instance, if I know exactly what ticket I want and simply want to buy it with a contactless card. But there are people, and situations, where that doesn’t work.
For example, on unusual journeys, I might want a human I can trust - and a machine will never do this - to advise, “I’ll sell you the Peak Single to X, and then you can get an Off-Peak Return back from Y. It’ll save you £9.”
O’Connor: Machines are getting better at this.
Sutherland: Sure, but there’s no incentive to make them better. It’s down to the decency of the human being.
That word – ‘incentive’ – is a crucial variable in this dynamic.
To shoehorn this idea into our PPC scope, it is an interesting illustration of exactly why the biggest problem with maximally trusting Google’s algorithms to guide our campaigns, is not that they’re not good enough at their job.
It’s what job exactly they are designed to accomplish.
Google isn’t incentivised to bake everything we would want, into its algorithms.
If Google could find a particular algorithmic tweak such that our campaigns would reach profitability a little quicker – and with a little less ‘exploratory’ spend – would they implement it? From our point of view it’s a no-brainer. From Google’s, it’s – at the very least – a more nuanced decision.
As Mike Rhodes puts it in a recent LinkedIn post:
“Google will never tell you to spend less money on ads! Even if that’s exactly what would make you more profitable.”
On top of that question mark around motivation, there’s also the practicality.
A viable algorithm (or small set of algorithms) designed to work across all accounts couldn’t factor in everything that mattered to us individually, even if Google wanted it to.
As with the ticket machine – our advertising endeavours often feature non-standard circumstances and sub-objectives – known or unknown.
Individual variances don’t make for good rules, but they are important – and they call precisely for the kinds of manual fine-tuning dials that have been, on the whole, decreasing in Google Ads (with some notable recent exceptions last year that go a long way towards redressing the balance in PMax).
Another interesting parallel emerges as the conversation continues:
O’Connor: (Machines are getting better at this) Like split ticketing: you can get those deals now.
Sutherland: Exactly, but those aren’t done by the railway companies. They’re done by intermediaries like Trainline.
So it is for us and PPC. Third-party tools like Opteo and Adpulse (my two favourites) focus on systematically identifying wasteful spend.
Wouldn’t it be nice if Google’s native recommendations did the same?
