Advertisement

SKIP ADVERTISEMENT

State of the Art

The Online Ad Industry Is Undergoing Self-Reflection. That’s Good News.

Credit...Doug Chayka

Digital ads are easy to hate.

Sure, they support much of today’s media and technology business. They give us social networks and search engines and — oh, for the love of Pete, can you just let me watch my video already? Why are you making me sit through 30 seconds of some guy trying to sell me boat insurance?

What I’m saying is, people tend to fly off the handle when they contemplate the modern advertising industry.

This irrationality makes it difficult to talk soberly about what’s really going on when the business of ads makes the news. It’s especially difficult to take the nuanced position when you’re talking about “programmatic ads” — ads that are purchased and placed according to algorithms that determine their relevance to you and that sometimes appear to follow you across your digital meanderings.

These ads have lately come under fire. The Wall Street Journal found that YouTube’s advertising engine was placing ads from large brands on racist videos. In response, several large advertisers pulled their ads. Meanwhile, an experiment by JPMorgan Chase called into question the effectiveness of programmatic ads. The company significantly reduced the number of ads it shows, but didn’t see a decline in response from users.

But let’s take a deep breath before condemning programmatic ads.

On the one hand, there are some clear problems with how programmatic ads are placed. The industry is rife with complexity. This type of advertising is also quite new, so a lot of the machinery that runs the ad market is still in the works.

But these problems are also fixable and should not obscure a larger truth: Even though they are far from perfect, in many ways programmatic ads are creating a more efficient advertising market. And given that advertising pays for nearly the entirety of what we see and do online, the upside of all the hand-wringing is that we are now examining how all of that money gets spent — a process that should lead to better ads, and better media, too.

To understand what’s at stake in the ad industry’s shift toward programmatic ads, we need to first traipse through a brief history of media and advertising.

Once upon a time the ad business was dominated by size. Big companies with lots of money needed a way to convince people to buy their stuff. Luckily the people were easy to find: They were all watching or reading one of a handful of media offerings — three TV networks, some big glossy magazines and one or two newspapers in every town.

So it was all pretty simple. The companies paid men in Manhattan to come up with catchy songs to get people to buy soda and soap and razor blades. The men got to drink martinis at work, so few had any reason to complain.

But glamorous as it was, the ad business wasn’t the best deal for the companies that were paying for the ads. The central problem was the conflation of audience and media outlet: When it ran an ad for razors, Gillette would have preferred to show the spot just to men who shaved regularly. But you couldn’t target men who shaved regularly.

“It wasn’t possible for us to be certain that we were reaching that audience, so we used the content of certain programming to define that audience,” said Brian Lesser, the chief executive of GroupM, a division of the advertising giant WPP. In other words, instead of targeting men, they’d run ads on shows they thought men liked to watch — a good enough solution, except for all the women and non-shavers who were also watching.

Digital advertising fundamentally altered this model. Through profiling, now ad companies know — or, at least, aim to know — exactly who is reading a certain site or watching a certain video. So instead of buying ads tied to a certain piece of content, companies can buy ads targeted exactly to an audience.

“Now we can get down to the level of an individual user and we can be certain that we’re targeting an ad to the same user across multiple devices,” Mr. Lesser said.

But it’s deeper than that. Ad companies don’t just know the user, but they also know the user’s context — for instance, whether you’re at work or at home, or whether you’re in the mood for shopping or not. All of this comes together in a real-time calculation as you wander around the digital world, from app to website to social feed. The computers are watching what you do and deciding which ads to serve you when. Often the ads are sold dynamically in an auction — different companies offer to pay different amounts to get your attention at different times.

There are some obvious downsides to this model. It relies on the profiling of users, which makes many people uncomfortable, even if the ad companies say that they do all of this anonymously and without invading your privacy.

It also raises an issue for brands. One consequence of this model is that it pays for a lot of content that wouldn’t have been funded under the old model — now a teenager can attract a few million followers on YouTube, sign up for the company’s revenue-sharing program and make money from all of the programmatic sponsors.

That’s a mixed blessing. It turns out there’s a lot of crazy stuff on the internet. Some of it is popular, even if it’s racist or otherwise objectionable. So suddenly a whole bunch of brands may be funding content that, in the old days of human-powered ad buying, they’d never in a million years have ventured anywhere near.

The best way to think of this is as a supply-chain problem. Big, complicated industries that sell commodity goods often get tripped up on sourcing: you don’t know who made your clothes, whether your diamonds are funding war or whether your shrimp may be complicit in slavery. The way to solve these externalities is through investigation and better documentation — to follow the money deep in the supply chain, to figure out who’s ultimately getting it.

Something like that is now happening in the programmatic ad industry. It’s rare to talk to someone in advertising who will tell you that things are working perfectly. They admit that there are holes in the system: Computers aren’t yet great at figuring out what’s racist and what’s not, or what’s fake news and what isn’t.

But they’re getting better, and the more activists push platforms like YouTube on this issue, the more incentive the platforms have to keep improving things.

“The nature of the internet is that it is fragmented,” said Ari Paparo, the chief executive of Beeswax, an advertising technology company. “You can’t buy over martinis everywhere you want to reach those customers. It just doesn’t work that way. There’s no amount of wishing that’s going to make the old days come back.”

Email: farhad.manjoo@nytimes.com; Twitter: @fmanjoo

A version of this article appears in print on  , Section B, Page 1 of the New York edition with the headline: Targeted Ads Are Annoying. But They’re Worth Saving.. Order Reprints | Today’s Paper | Subscribe

Advertisement

SKIP ADVERTISEMENT