I wonder how close ad markets are to zero sum. In the limited cases where ads are good for an individual firm, being close to zero sum would make ads a drain on the industry. In the article, they talk about how in most experiments, the most of the people who click your ad would have clicked your link anyways without the ad. In a really important footnote they talk about how this is crazy with a hugely important exception: if you don't bid, your competitor will. Like amazon bidding on searches for "walmart."

The effect is obvious and obviously wasteful for brand name searches, but it must extend to other searches as well. I search for "levi's" and your ad isn't going to make me materially more likely to buy jeans, but it will make me more likely to buy them from Levi's. Or, and here's the point, if I just search "jeans," your ad isn't going to make me materially more likely to buy jeans. But you have to bid or else I won't buy your jeans. It strikes me as just as wasteful as Levi's having to bid on the "levi's" search term. The money that was going to go towards the jeans industry still goes there, except that now Google and Facebook get a cut and the jeans industry's margin shrinks a little.

This article does a brilliant job convincing me that most ads aren't incremental for a firm because of selection effects, which convinces me that most ads that are incremental for the firm aren't incremental to the market. It seems so wasteful.