select all

The Best Way to Ensure Automated Ads Are Appropriate Is to Make an Intern Click Through 12,000 Websites

Photo: Bloomberg via Getty Images

Most companies that buy ads online rarely do so by selecting sites themselves, and choosing what their ads actually appear alongside. They rely on automated, or programmatic, systems from companies like Google and Facebook, which — in a matter of milliseconds — decide which of their clients to put in the ad unit whenever a page is loaded.

There are many upsides to these systems. For advertisers, it makes it simpler to target an intended audience, “men over 40,” “women searching for beauty products,” and so on. For small operations and individuals, it lets them display ads and collect ad revenue without having to sell the space themselves.

But the programmatic-ad industry is having a rough few weeks. On YouTube, according to The Wall Street Journal, ads for top-tier brands like Coca-Cola and Microsoft were appearing alongside racist and anti-Semitic videos, causing companies to pull their ads.

From the Journal:

Asked about the Journal’s finding that their ads were still appearing with such content on YouTube as of Thursday night, Coca-Cola, PepsiCo Inc., Wal-Mart Stores Inc. and Dish Network Corp. said Friday they were suspending spending on all Google advertising except targeted search ads. Starbucks Corp. and General Motors Co. said they were pulling their ads from YouTube. FX Networks, part of 21st Century Fox Inc., said it was suspending all advertising spending on Google, including search ads and YouTube.

This particular dustup comes on the heels of months of criticism over how programmatic-ad platforms allow sites specializing in hoaxes and fake news to draw in tons of revenue by perpetuating misinformation. Google and Facebook have since said that using their ads on sites distributing fake news violates their policies.

Now, there are growing questions about how effective these ads even are. Probably the most popular online-advertising metric is an “impression,” when an ad is seen by a user. An audit from JPMorgan found little change in the cost of an impression, even when it reduced the number of sites its ads appeared on from 400,000 to 5,000. The culling, according to the New York Times, was the work of one unfortunate intern.

Of the 400,000 web addresses JPMorgan’s ads showed up on in a recent 30-day period, said Ms. Lemkau, only 12,000, or 3 percent, led to activity beyond an impression. An intern then manually clicked on each of those addresses to ensure that the websites were ones the company wanted to advertise on. About 7,000 of them were not, winnowing the group to 5,000.

The question yet unaddressed is how these practices, if adopted widely, will affect smaller sites and channels that rely on large corporate advertisers casting a wide net. If those ads vanish, so does the only revenue option for many smaller sites.

Intern Horror Story: Forced to Click Through 12,000 Sites