Contextual advertising isn’t the privacy-safe panacea everyone thinks it is.
As user-level IDs diminish on the web and in apps, publishers and ad tech companies are fighting over what “contextual advertising” even means and who has the right to serve contextual placements.
The purpose of contextual advertising is to target elements of the media, like words, images and moods, as opposed to targeting a user or device. But that definition has crept further down the spectrum toward user-level tracking.
Now, some ad tech companies are tracking browser and page-level data, device data, IP address, location data and whatever other info they can get their hands on to model the potential user, framing it as “contextual 2.0.”
Five years ago, though, those same companies may have unabashedly called that fingerprinting.
Publishers, DSPs and other ad tech companies are converging on contextual because it offers “survivable signals” in the new world of mobile, web and CTV, as it expands with changing times, said Mario Diez, CEO of one of the old standard-bearers in the category, Peer39.
Publisher pushback
Publishers, who intuitively should benefit by the growing importance of contextual advertising, are actually disadvantaged.
They have in-depth knowledge of their content, authors and audience, if, say, readers have a propensity to share links or certain stories are likely to generate affiliate traffic. Publisher-based contextual solutions are inherently better, according to Paul Bannister, CafeMedia’s chief strategy officer. Publishers understand the author, have their own meta-data on stories and actually want ads to be a good fit for their pages.
But those aren’t contextual signals in the sense of targeting language or mood on a page. It is “contextually derived” data and can be a valuable way to increase ad value, said Scott Messer, a publisher consultant.
For example, if a site has a mortgage calculator, that publisher might create an audience segment that’s “in-market for a mortgage,” he said. That‘s an example of taking a contextual signal and fusing it with behavioral data (i.e., tracking a user around the web) to retarget readers.
“The publisher advantage is speed, accuracy and depth,” he said. A publisher that posts dozens, if not hundreds, of articles a day understands how each fits into news trends in their coverage. A scraper just can’t keep up with that pace.
But the disadvantage is greater still. There’s nothing publishers can do about Google, Grapeshot or even their own verification vendors like Integral Ad Science and DoubleVerify scraping pages to create contextual products that span the open web.
What’s most frustrating is that advertisers themselves also don’t effectively use contextual solutions, Bannister said. Blunt-force keyword and category blocking is still the norm.
Advertisers are very sophisticated about keywords when it comes to SEO, Messer said. They expand definitions of words to find new ways to reach similar audiences and are constantly feeding data into analytical models. It puts to shame the approach with contextual where any pages with the word “war” or “politics” might be discarded without a backward glance, and there’s little effort to sharpen the strategy moving forward.
Contextual keyword tactics that make sense for one campaign or moment in time pile up like a clogged artery, said Josef Najm, global lead for programmatic sales and partnerships at Reuters. If a plane crashes or a cruise ship is on fire, those terms become contextual signals to avoid.
When the Dobbs case overturning Roe v. Wade was decided last summer, for example, advertisers turned on keyword blocks for terms like “Supreme Court,” “abortion” and even medical terms like “OBGYN” or “women’s health.”
Those blacklist terms don’t fade, Najm said; they accrue.
Major political sites like Reuters, BBC and The New York Times are still burdened by keyword blocklists from the 2016 election, when everyone was fed up with political news, he said. Contextual solutions that were pitched as a way to avoid advertising on Breitbart or other conservative news sites are still filtering ads away from valuable reporting, six years later.
Even in the open web, most legitimate inventory is consolidated among the top 50 or so publishers and media sellers. “That’s a finite number that an advertiser can work with and actually understand their taxonomy,” Bannister said.
And brands don’t appreciate how easy it would be for them to operate their own contextual framework, he added.
Rather than prioritizing scale for scale’s sake, advertisers could work directly with publishers on a more one-to-one understanding of context for that environment.
Contextualized vendors
DSPs and new contextual players like IAS and DoubleVerify promise contextual targeting with nothing but the click of a button.
Some advertisers may think IAS and DoubleVerify’s contextual solutions are based on data licensed from their publisher networks, or that publishers agree to allow contextual targeting across sites by opting into IAS or DoubleVerify. But these verification providers scrape the web like everybody else and make contextual judgements across sites without relying on publishers for input.
Anyone can scrape the internet and parse basic language, but most contextual solutions are still using third-party IDs to target and segment audiences, said Geoffroy Martin, CEO of the ID-free targeting company Ogury.
An overall trend in online advertising is to consolidate with a trusted partner, rather than “piecemealing” together a solution for many publishers and point solutions.
The advantage — and challenge — for contextual providers like Peer39 or Oracle-owned Grapeshot is to apply their contextual offerings across publishers, DSPs or SSPs with their own contextual offerings, Diez said. “From the buy side, what’s most important is consistency and a normalized view [of the media].”
Verification and suitability
Getting into contextual targeting was a natural next step for IAS, said SVP of product Craig Ziegler.
“Those capabilities that started as brand safety and suitability got more sophisticated and have evolved into contextual targeting,” he said. In other words, IAS went from vetting inventory for whether an advertiser shouldn’t appear to judging where an advertiser should serve an ad.
Brand safety, ad verification (like whether an ad was viewable), attention metrics and contextual targeting are all part of the same way marketers judge an impression, he said, so it makes sense for those capabilities to sit in one platform.
All those capabilities “require contextual understanding,” he said.
Brand safety is purely about avoiding a publisher, URL or certain semantics on a page, Ziegler said. But he added that contextual advertising is about identifying opportunities across the web, not just understanding what’s on a page.
“Targeting is seeking out certain contexts to be in deliberately,” he said. “And that’s generally the distinction that’s most popular and the one that that we go by when we talk to customers.”
