If I were buying ads (I've made some micro purchases in the past), I'd also understand that the users of the small, niche website are also likely using the larger websites, too.
So the time in researching what small websites exist for my market (say dog owners), the ad cost per buy of making many small buys, the administrative overhead related to maintaining many small accounts (accounting, dealing with many sales people, many platforms to administer, etc) vs. buying ads in greater bulk with a few content providers like the NYT or others alongside pet related stories (or other compatible content) where I'm likely to get most of the niche site users and others that are also in my market that don't use those sites.... seems kinda no-brainer.
I also expect data aggregation to continue, even without a focus on "personalized" or "one-to-one" behavioral data. For example the NYT times would likely want to be able to show advertisers that those that read dog articles also disproportionately read articles about outdoors activities... they can produce those metrics within the course of normal operations, I would think, possible even without employing things like cookies for the purpose... naturally the metrics being captured today do have a market value beyond just allowing the collector what ads to show and my suggestion doesn't replace that path, but I expect for just the purpose of targeting ads that's probably more useful anyway.
So the time in researching what small websites exist for my market (say dog owners), the ad cost per buy of making many small buys, the administrative overhead related to maintaining many small accounts (accounting, dealing with many sales people, many platforms to administer, etc) vs. buying ads in greater bulk with a few content providers like the NYT or others alongside pet related stories (or other compatible content) where I'm likely to get most of the niche site users and others that are also in my market that don't use those sites.... seems kinda no-brainer.
I also expect data aggregation to continue, even without a focus on "personalized" or "one-to-one" behavioral data. For example the NYT times would likely want to be able to show advertisers that those that read dog articles also disproportionately read articles about outdoors activities... they can produce those metrics within the course of normal operations, I would think, possible even without employing things like cookies for the purpose... naturally the metrics being captured today do have a market value beyond just allowing the collector what ads to show and my suggestion doesn't replace that path, but I expect for just the purpose of targeting ads that's probably more useful anyway.