[jonreem]

In your analysis you shouldn’t forget that advertisers don’t consider all eyeballs equal either, and are willing to pay far more for ads shown to high income users. The difference in effect might not be as large as you’re making it out to be.

[scarejunba]

There is sufficient differentiation in advertising strategies that this can be generally true but also be misleading. You want people who you're targeting to switch from non-spending to spending. So if you've got a new brand of diapers, you're less likely to be targeting Mr. Rich Dude than you are pregnant mum with only minor income category stuff.

That's fairly different from where only people with disposable income get to see things because only they are able to pay for them. If a lower-middle-class recent mother can view a maths video after viewing an ad for a specific brand of diapers (that sets out to get her to switch from a different brand), that's a better outcome than if she can't because she's not a middle- or upper-middle-class person.

I suspect the difference is actually fairly high in practice. Judging by some ad campaigns that are facilitated through our middle-ware, it's not all Maseratis and fancy wines. Quite a bit is for the bottom 80%.

[kjeetgill]

Have you worked in advertising? It's nothing nearly so simplistic, but not so complicated either.

It's not about capturing the highest income audience its capturing the largest audience in the income demographic of your product.

Think something like (amount a customer will spend) * (% of demo willing to spend at all) * (size of income demo). Naturally more people are less rich and the first two terms are highly product dependent.

Stereotyping a bit, McDonalds isn't going to be making big inroads showing all of their ads in opera houses and Lamborghini is not making any sales advertising at Goodwill.

[retsibsi]

This still implies that higher-income eyeballs will be more valuable, doesn't it? McDonald's might not value rich people so highly, but Lamborghini will be willing to pay 10,000x more per customer gained. Of course the details could confound this (supply and demand for each category of consumer, conversion rates, etc.), but it seems like a reasonable default assumption.