Can Contextual Targeting Replace Third-Party Cookies?

Contextual targeting is all the rage – and no, it’s not 1998.

It’s 2019, and the industry is anxiously casting around for an alternative to third-party cookies as the browsers clamp down on cross-site tracking and regulatory pressure ratchets up in the United States and around the world.

A shift away from audience-based buying is inevitable, said Max Jaffe, managing partner and programmatic practice lead at GroupM.

“You could argue that over the years, the pendulum swung too much toward audience-based buying and third-party data – and now it’s starting to swing back,” he said.

Reactionary or for real?

The California Consumer Privacy Act, which goes into effect on Jan. 1, 2020, has a lot to do with the growing interest in contextual stateside, said Jay Friedman, president of programmatic agency Goodway Group.

If cookies are no longer viable, keywords are a quick fix.

Although targeting based on authenticated traffic is a more viable alternative to cookie-based behavioral targeting, it’s difficult to implement, and so people will “flock to contextual out of fear or a lack of other ideas,” Friedman said.

But the flocking hasn’t started yet. Goodway Group at least isn’t seeing increased demand for contextual from its advertiser clients – although that’s not surprising. Advertisers often care when they have to care, and since it’s still possible to target with cookies, “that’s what advertisers are going to do while they can,” Friedman said.

Contextual grows up

Meanwhile, companies that focus on contextual targeting, such as GumGum, the newly PE-backed Peer39 and Zefr, are revving up for when the opportunity ripens, while others, like Oracle Data Cloud, are actively pulling away from risky third-party data to knuckle down on contextual.

Publishers, such as The Washington Post, and technology providers like video ad platform Unruly and DoubleVerify, are also investing in contextual technology.

One of the main motivators behind DoubleVerify’s acquisition of Finnish contextual startup Leiki earlier this year, for example, was a bet on the prediction that “contextual is going to be one of the next big things when it comes to effective online ad buying,” said Roy Rosenfeld, SVP of product development at DoubleVerify.

And as far as tactics go, said GumGum’s CTO, Ken Weiner, contextual “isn’t just about looking for simple keywords on a page anymore – there’s a lot more on offer today, and it’s getting more and more sophisticated.”

Still, the perception persists that contextual targeting is a less-sophisticated blast from the past that’s only seeing a resurgence thanks to uncertainty about the future. What’s so interesting about placing ads for golf balls on a golf website?

“But there’s actually a lot you can do beyond keywords – and the fact is that even keywords are an extremely strong signal,” said Kurt Kratchman, SVP of product development at Oracle Data Cloud and former COO of Grapeshot. He claims the brands and agencies ODC works with are intending to increase their spend on contextual.

ODC, for example, categorizes every word on roughly 15 billion unique pages every month to connect page engagement with audiences and create custom segments. Advertisers can also create dynamic keyword targeting lists in 33 languages that automatically get updated based on what’s likely to achieve higher performance and better reach, Kratchman said.

What you can’t do with contextual

But contextual is not going to “save the programmatic ecosystem,” because it’s hard to measure, said Steve Silvers, VP of product management for the Neustar Identity DMP, speaking during a New York Advertising Week panel in September.

Without cookies, device IDs or other identifiers, measurement gets murky, and for more spend to go toward contextual, advertisers are going to have to accept less granularity, said Emily Roe, associate director of media services at Dentsu-owned Merkle.

The industry needs to develop benchmarks for contextual so advertisers can determine whether they’re getting a bang for their bucks.

Frequency capping and attribution will also pose a challenge, or at least they won’t operate the same way they do in a cookied world, said Neal Richter, chief scientist at SpotX.

“These things are probably going to get a little … fuzzier,” Richter said. “Frequency caps will have to be more of a target than a constraint, and there will be a move away from the false precision of trying to track independent users from end to end in an event stream.”

For those reasons, contextual might seem less scaled and less performant than other methods, but it’s “still a valid tool in the overall portfolio,” said Bill Simmons, CTO and co-founder of dataxu.

“It’s time to give contextual a fair shake, and that’s going to happen if we see people starting to opt out,” Simmons said.

But for contextual to get a fair shake, companies need to invest in the tech, and contextual targeting technology overall is still hugely underfunded compared with audience-based tech, said Ratko Vidakovic, founder of ad tech consultancy AdProfs.

“The result of that is clearly illustrated by the fact that there are only a handful of contextual targeting companies out there,” Vidakovic said.

And the current zeitgeist is why so many companies are trying to position themselves as the go-to solution for when cookies officially crumble.

“We’re betting on 2020 as the year of contextual,” said Richard Raddon, CEO and co-founder of Zefr. “There’s a time coming, and soon, when people will realize they can’t rely on third-party segments anyway – and that maybe they never should have.”

But is contextual worth it?

If 2020 is the year of contextual, what happens to publisher CPMs?

Today, behaviorally targeted ads usually garner somewhere between 50 cents and $2 or $3, depending on the specificity of the segments, whereas contextual signals are in the 10- to 30-cent range on a CPM basis, said Peer39 COO Alex White.

But in a perhaps-not-so-hypothetical world in which contextual becomes the dominant method for targeting, CPMs should eventually level out – as long as advertisers don’t move all of their budgets to Facebook, Google and Amazon, said GumGum CTO Weiner. 

“If people can’t deal with no audience targeting and they move to the walled gardens, CPMs will go down outside of the walled gardens,” he said.

But the death of cookies doesn’t mean the death of audience targeting, said Goodway Group’s Friedman.

“Amazon and AdSense and Google’s DSP – they’re all built into the open web,” he said. “If you’re a publisher, you might start seeing AdEx bids rise up while every other SSP starts to see their bids measurably deflate – that could happen.”

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!


  1. Unfortunately, this is yet another article that casually conflates 3rd party cookies with 3rd party data, and seems to imply that CCPA will have a catastrophic impact on audience targeting — but doesn’t explain why that would be. The “Do Not Sell” provisions of CCPA apply to data collected by specific entities, for example a publisher site. So a consumer might visit a publisher site, click “Do Not Sell”, and that would apply to data collected by that publisher. It DOES NOT apply to data previously collected in a proper manner from other sources. It would be great if these important distinctions were pointed-out.

  2. You’re right that CCPA will apply to each publisher site/data collector individually, and be confined to each entity. But with the application of CCPA translating to an “opt-out” system (vs. opt-in/”say yes to these cookies” a la GDPR), studies show consumers are very likely to hop on the “do not track me, bro” train. This will likely have a cascading effect of tracking users across many publishers at its onset; but as publishers determine what content is free vs. requiring a paywall/login, we ought to see a bounce back after an initial period.

    Proper collection will always inherently be compliant. But it puts a serious strain on using non-publisher data (like a lot of data aggregators available in DMPs) for the sake of targeting, which will likely have a strong impact on scaling programmatically using data. I foresee publishers having a “come back” era, which may or may not artificially inflate a universe of more walled gardens. While what you’re saying is valid, I interpret this article to be more about scaling with data-centric targeting vs. the complete unraveling of the fiat currency of data collection altogether.