[samizdis]

> A decade ago, economists at eBay, including a former professor of economics at UC Berkeley, proposed an experiment to see how effective the company’s marketing was. All eBay search ads on Google were stopped for three months in a third of the US. There was only a small difference in sales. The economists concluded that the people clicking on search ads were the ones already planning to buy something on the site. eBay was wasting millions of dollars. Other companies may come to similar conclusions. Airbnb slashed its digital marketing budget last year in response to the pandemic and says that it saw no change in online traffic. ...

> The dirtiest secret about targeted advertising could be that it doesn’t work.

[jfengel]

The case for any advertising by big companies has always been week. You already know eBay and AirBnB. They're already so popular that they're going to be the organic first search result. They would be buying ads mainly to prevent other companies from competing with them.

Also, these are companies that basically everybody needs. Everybody buys stuff and stays in hotels, and these two are the main competitors for the Internetty versions of those things. Even if you don't use them, you'll have heard of them.

The use of targeted advertising is when you're trying to target. You're a Yemeni restaurant who doesn't expect clientele to travel more than 5 miles, and people who have at least a stab at adventurous eating. You're a dentist who really nailed their new dental office manager software and don't expect to sell it to 12-year-olds, or auto mechanics. You're a theater troupe competing for the entertainment dollars with movies, video games, etc. and want to see if you can put your show in front of people who like going to theaters -- or who might be curious about it.

The latter is my case. We throw $20 at ads, and we get maybe 3 or 4 people who come that wouldn't have otherwise. That about breaks even, maybe, but even better one may subscribe to our notices by email or Facebook.

It just seems dumb to go looking at the people who target everybody and then declare that target ads don't work for actual targeting. That may or may not work; it's aggravatingly hard to measure. But you could at least go looking at the cases that might be useful.