>Who would opt-in to those?

The OP's approach doesn't depend on people opting into legal ads. You can use ML/regressions/whatever to extrapolate the best people to show which type of legal ads based on their interests, even if those interests do not include "Legal". There doesn't even need to be an opt-out option at all. I don't think consumers would mind, since opt-outs reduce the service's advertising revenue which in turn increases the cost of the service. Almost no one wants to pay more money.

>I don't think consumers would mind, since opt-outs reduce the service's advertising revenue which in turn increases the cost of the service. Almost no one wants to pay more money.

Your natural learning just failed. Congratulations. Can't wait to see the Machine learned version you want to force on people completely unawares of what you're doing.

Privacy bloody means something. Until you start respecting people's desire to be left well enough alone, and not have their attention imposed on by obnoxious advertising, all you are going to door ensure that the wave of sentiment against the practice grows more and more severe.

So we’re not just anti-tracking now, but also completely anti-advertising? How else do you think free services will get the revenue they need? Do you think all of them will be able to survive solely on patron revenue and purchases? Some surely can, but you’re making a universal argument here.

Wherever advertising is necessary to keep a website running and up to date, being anti-advertising is akin to being anti-working to get your money or anti-walking to get to your destination.

> You can use ML/regressions/whatever

But do you not feed that with all sorts of data you aggregate about the user?

When a user volunteers granular data about their interests, you can use pre-existing correlates between those interests and purchasing probabilities. Those correlates can be derived from exogenous data generated from a subset of users with the same interests (who voluntarily accept more tracking for free premium features), experiments, educated estimates from experts, etc...

The above approach might not be perfect, but I would compare it to the recent trend towards organic food. Organic food is simply costlier and less efficient to produce. Nonetheless, a growing number of people wouldn't do without it. They only visit stores with organic meat and produce. Websites that completely eschew tracking and third party tracker data can be the "Whole Foods" of websites.

You can infer a lot about which sort of stories people read especially for news papers - and also their geo location.

Basic ML clustering - its not perfect.

If people don't check the "legal" or "insurance" boxes, then they wouldn't be the most profitable any longer, something else would be. And it would be something people are proactively more interested in. Besides, if you're in a situation where you might need a lawyer or you might need a type of insurance, you might actually go check that box, even if you uncheck it again once you've found one.

If the top categories apply for most people by default, then there's no reason to have people opt-in or track them.

The general categories apply to everyone. The profitable subcategories are very specific, especially based on location and income.

got a source on that claim?

The adsense niches which pay the most:

https://alejandrorioja.com/high-cpc-adsense-keywords/

You also see it in Google keyword prices:

https://blog.hubspot.com/marketing/most-expensive-keywords-g...

The most profitable keywords are in industries where there is a high lifetime customer cost. Most of those are mundane because the less you think about it, the longer you stay as a customer.

I would have said the more ambulance chasing legal queries and plumbers 15 years ago when I did a bit of AdWords plumbing keywords in London where £45 to £50 quid

>> Adblocker detected! Please disable to enjoy my marketing guides.

Sure...