That’s sales, not marketing. You wouldn’t bother to create ads for a Da Vinci, as the pool of people with the means to buy one is so small and the process of buying one is so convoluted. See also jet engines.
> You want to make people buy your brand of beer? Well you have to sell the story of it being generally accepted as a high quality brand.
The article is contesting that what you actually need is to make people aware of your brand at the moment they are about to consume one. The brand associations matter in this on some level, but are tangential. Stella Artois has completely different brand connotations in the US to the UK and EU for instance.
The ad that's just looking at people looking at it is brilliant, really a work of art on its own.
https://www.youtube.com/watch?v=jsEAJkTP0-M
The effort that went into selling that painting is one of the more fascinating stories of the last year.
"Before Christie’s sold Leonardo da Vinci’s Salvator Mundi in 2017, it staged high-profile public exhibitions of the painting around the world, and made an expensively produced video to showcase it – in effect, an ad. On the face of it, all this activity was a flagrant waste of money: almost nobody who consumes the marketing for a da Vinci is in the market for it. Christie’s salespeople knew all the potential buyers personally. They can, and did, visit them in the privacy of their penthouses.
"Yet Christie’s realised that those buyers would pay an extra few million for the privilege of owning a painting that was “iconic”. This isn’t some quirk of billionaire art collectors; it’s human nature. We value things more highly when we know that others value them."
Similarly, Heineken is very much worth trying in the Netherlands. I’d never drink it here in the US.
You can see a curious parallel in the death of psychoanalysis and the rise of evidence-based psychology and CBT but somehow depression is still not getting better. I don't believe Man is as simple as something to be fully captured by psychographic profiling. Articles like this are precisely the symptoms of the evergrowing decline in symbolic efficiency of the surveillance capitalism regime.
It doesn’t always fail, but it fails more than rolling a die.
Though if it fails more often “than a roll of a die,” doesn’t that hint at some predictive power? Maybe I’m misunderstanding.
Imagine we scored you as a likely alcoholic because people with a case history like yours indicated a high risk of alcoholism. So we start taking preemptive measures and the entire system grinds into gear, partly trying to prevent you from falling to the bottle but also preparing for if you do. You really wouldn't want to interact with a public system that's put the wrong label on you, trust me. So we have to be more than a little sure we get things right.
You can probably use ML to predict general trends, but I don't have much data on it, because we use traditional analysts in the public sector. Not because they are great at it, but because they're there. We've been predicting general trends since before computer technology, and the people who work in those departments typically aren't very tech savy. It would be a tremendous cost of resources to make them use ML, or to replace them with data scientists.
I guess in the future those analytics departments may increasingly rely on ML, but to be completely honest, I have no idea if ML or traditional analytics is better at predicting the outcome. The fact that we haven't started using ML, haven't hired a single data scientist or signed a single contractor on the area probably speaks for itself though, if it was truly valuable, I suspect we'd have it by now.
By comparison, RPA didn't really have a great start in the public sector, but today it's absolutely everywhere because it turned out to be really valuable.
It's not like we don't use ML at all though. We do, for stuff like sorting through four million case files looking for specific errors. Something that takes tens of workers months of work to do, but can be done in a few hours when you rent enough hardware in the cloud. We just don't use ML for any sort of analytics or prediction.
"Imagine we scored you as a likely alcoholic because people with a case history like yours indicated a high risk of alcoholism. So we start taking preemptive measures and the entire system grinds into gear, partly trying to prevent you from falling to the bottle but also preparing for if you do. You really wouldn't want to interact with a public system that's put the wrong label on you, trust me."
That's exactly the sort of specific example that is missing from the stories about ML doing things like encoding existing biases.
its headline
You want to sell a da Vinci? Well then you have to make private meetings with potential clients, and, you have to create ads that sell it as iconic.
You want to make people buy your brand of beer? Well you have to sell the story of it being generally accepted as a high quality brand.
You want to sell luxury? You have to sell timelessness and trust.
Aren’t those things exactly what Don Draper would have done?
In general the article skimps over how ads are actually made, and talks mainly in how they are delivered, but that has very little to do with the death of Don Draper. Because even if you’re not always shooting ads at everyone, you still have to tell stories that set you a part from everyone else.
It’s only valid point in the subject seem to be that advertisers are no longer in the front of business magazines. Ironically, neither are car manufacturers, except for one, Musk. Who coincidentally put one of his cars in space in a very spectacular PR event that I personally doubt he decided upon because an algorithm told him it would be popular.