The web magazine Slate has earned a reputation for counterintuitive arguments. But even by Slate’s standards a recent article about Internet advertising was unusually provocative.
The article is headlined “We Have No Idea if Online Ads Work” and that pretty much sums up the argument. Considering the hyperbolic claim, it’s tempting to dismiss the article outright. But the author, Jordan Weissmann is Slate’s senior business and economics correspondent. And if he’s ultimately dead wrong, his attack on Internet advertising actually ends up making a great case for programmatic buying.
Before going into what the article gets wrong, let’s start with what it gets right. As Weissmann correctly points out, untangling cause and effect can sometimes be tricky when it comes to advertising –not just Internet advertising but all advertising. Weissmann cites the example of a company running a campaign during Christmas and then having no way of knowing if the subsequent sales increase resulted from the campaign or from the simple fact that it was Christmas.
This is all fair enough. Weissmann then goes on to suggest that, as with any science, the best (and only) way to untangle cause and effect is through controlled experiments. Agreed. But it’s at this point that Weissmann’s argument grows more flimsy. It turns out that his grand thesis that all of Internet advertising is in doubt is based largely on two experiments with search ads.
The first experiment found that one large company was wasting money by taking ads out against searches for the company’s own name, a point that seems so obvious it hardly needs testing.
The second experiment was more interesting. The researchers removed the same company’s search ads for certain keywords in some regions but left the ads up in others. As Weissmann explains, the ads didn’t have much of an impact on the company’s frequent users, but they did succeed in bringing in new customers.
As one of the researches told Weissmann, the ads “were extremely profitable,” when it came to new users. The problem was that the profitability was outweighed by the waste of serving ads to people who would have ended up at the company’s site even in the absence of the ads.
So, it turns out the issue wasn’t that the ads didn’t work but rather that they were often being shown to the wrong people. It’s hard to think of a better call for more programmatic advertising. The whole point of programmatic, after all, is gathering as much data as possible to target only those users who are worth targeting. The experiment is not an indictment of Internet advertising but of misguided Internet advertising.
The rest of Weissmann’s article focuses on a report by Randall Lewis of Google and Justin Rao of Microsoft that found that it can be incredibly difficult to figure out the precise ROI of a given digital campaign. That’s true enough (if not exactly breaking news). But it’s also true that we have countless cases where everyone agrees that an online campaign worked because of survey data or because of extremely strong correlations between the campaign’s run dates and sales.
It’s no wonder, then, that Lewis, one of the authors of the report, told Weissmann that “online ads absolutely work.” In fact, Lewis has written other papers that demonstrate that users exposed to display ads on Yahoo make more purchases in stores.
In other words, Lewis’s research shows something that programmatic marketers have been saying for a long time: If you really want to understand how display ads work on the Internet, forget about clicks and look at the number of people who make a purchase after being exposed to the ad– A.K.A. view-through attribution.
No, view-through attribution is not perfect. Some of the people who were exposed to an ad might have made their purchases anyway. But the research of Lewis and others shows that the view-through effect is real — as does the fact that users will often begin to search for companies after seeing display ads that were never clicked.
Let’s also remember that with view-through attribution, we at least have a way of making a connection between seeing an ad and making a purchase. When it comes to traditional, off-line advertising, we don’t even have that. And yet just about every major company in the world continues to run off-line ads.
So, yes, some people in Internet advertising do exaggerate when making claims for their offerings. But we all can all safely say this: We have a lot more evidence for what works online than we do with any other form of advertising in history. The lesson isn’t to give up on Internet advertising; it’s to make sure you’re getting your message to the right people and measuring success in the right way.