What space exactly? Individual niches are not as interesting because you might think that the market for your calligraphy kit is a forum for Japanese speakers

What space exactly? Individual niches are not as interesting because you might think that the market for your calligraphy kit is a forum for Japanese speakers (10% conversion rate, 10k people largest forum) but Manga-watchers have a much better potential (2%, 5M). It’s easy to assume who your niche is, but almost every campaign is a surprise. This is why Custom Audience powers most ads. Therefore your market is a service that knows about many people’s niche interest, but is itself quite generic: that’s Wordpress·com (freemium, not interested in ads), reddit (not good at monetisation for a long time), twitter (same delay), GMail to some extend and much later Slack, Discord, maybe SnapChat and TikTok. There is a duopoly in ads platforms because there’s a duopoly in effective ideas that scale to any support and any advertiser: keyword (Google) and personalisation. Many platforms have tried to base ads on the content of the page, but it’s hard to find ads that pair well with the war in Darfur. Many YouTube creators get away with promoted content, but it is murky more often than not. The most successful “editorial” attempt is OutBrain and Taboola that are essentially auto-generated articles with super-optimised clic-bait-y garbage. You can imagine hosting that information yourself, edit it, expose it to ad slots in your browser with a protocol. Dave Winner certainly dreams of it. But for me, the proof that this could work was Facebook letting anyone edit their ad preference and information: for years, you could just tell Facebook that you were _not_ into cars, baby clothes or stamp collection. No one used it. It was linked on every ad; you were linked to it if you right-clic to say an ad wasn’t relevant; it was easily findable on the website and the app… No one touched it. So much so that improving it was impossible without traffic. Privacy advocate who claimed this needed to exist ignored my comments when I told them it already did. There was major internal support for it, but even VPs can’t argue against data. You can look at the history: Facebook was created in 2004; they immediately had ads, but they only became a significant player on targeting from 2008 onwards. At that point, ads had been around since 1995 with Viaweb, so more than 15 years; Google had been around for since 1998, so 10 years and the acquisition that created AdWords five years earlier in 2003. If an alternative direction was easy, there was time to build something. I’m not saying another reality isn’t possible — I dream of a protocol emerging from Facebook opening its preference embeddings, after realising that there’s a worthwhile ecosystem of apps to let people optimise it the way they want. But not enough people wanted it, while billions of people seem comfortable with delegating decisions and targeting entirely.