Super Bowl ads work, sometimes. Brain scans of 2006 Super Bowl viewers watching different commercials revealed dramatically different fMRI results for different ads, confirming what you already knew. Some Super Bowl ads are brain-tickling pieces of art. Some Super Bowl ads are toe-curlingly awful.

Even if many of the ads fail, there does appear to be something like a “Super Bowl bump” in the market. The stock prices of Super Bowl advertisers between 1996-2010 increased by $10-20 billion annually for the two weeks around the big game, according to research from the University of Wisconsin, Eau Claire. Another study found “significant abnormal net buying activity,” particularly among retail investors, up to 20 days after Super Bowl ads. This brief post-game spike didn’t necessarily last longer than a fortnight. But it suggests that, even if viewers don’t love all Super Bowl advertisements, investors do.

What about consumers? After all, Super Bowl ads aren’t art installations. For them to “work,” they must convert viewers into paying customers. The evidence here is mixed. In a new research paper, “Do Superbowl Ads Affect Brand Share?”, Wesley R. Hartmann of Stanford Business School and Daniel Klapper of Humboldt University Berlin studied Budweiser, Coca-Cola, and Pepsi commercials to determine if beer and soda companies have used the game to add market share. In both cases, the researchers were “unable to find an economically significant relationship between advertising and sales.” In fact, the rate of return was so negative, it called into question the entire premise that TV advertising works for well-known commodities.