Why do contextual ads fail

Google search sometimes gets it right, as do a few other Google services. But those experiences are rare. More often, the ads miss the mark. When I go to YouTube, I see ads for cars (which I'm not in the market for and have no interest in). Google's privacy policy lets it take what it learns about me via one Google product and use that information to serve up ads on another Google venue.

Advertising on Google Search and in Google Ads on Amazon and other websites mostly seems to promote things that I've looked at or already purchased. For example, if I buy a wallet, I see hundreds of ads for wallets for months afterward -- the one thing I definitely don't need.

But seeing that I was shopping for a wallet, then serving up ads based on that behavior is hardly sophisticated contextual advertising. Where does the endless list of personal data and signals go Google and Amazon both know what I read, what TV and movies I like, where I live, how old I am, my gender, my interests, my professional interactions, and so much more. What are they doing with that information It's clearly not doing me any good on the advertising front.

And it's not just advertising, but content, too. Google and Facebook algorithmically filter what you see in your Circle Streams or News Feed, respectively. They show you some of what your family and friends post, but not all of it. We're supposed to trust their algorithms to show us what we want, based on our personal data and activity. Yet in both cases, they fail miserably.

Everyone on Google+ has a Notifications view. In my experience, half of my notifications are relevant -- showing the kind of content I want -- and half of them couldn't possibly be relevant to anyone.