There is a particular cruelty to what is happening to digital journalism right now. News publishers are producing the most important, most widely read journalism of the past decade. Their traffic metrics — page views, unique visitors, time on site — are at or near all-time highs across the board. And yet the programmatic advertising revenue that sustains most of their operations is generating a fraction of what it would generate under normal circumstances, because the brand safety technology that advertisers have deployed is blocking their ads from appearing on virtually any page that mentions the word “coronavirus.”
The consequence is not an abstraction. It is the financial starvation of journalism at the moment when journalism’s value to the public is most apparent. And the mechanism causing it is not malicious — it is simply blunt, unreflective, and long overdue for a fundamental rethink.
How We Got Here
Brand safety as a commercial practice emerged from a series of high-profile incidents in which major brands found their ads placed adjacent to extremist content, graphic violence, or politically charged material they had not intended to support. The YouTube brand safety crisis of 2017, in which brands including major consumer packaged goods companies pulled spend after ads were found running alongside extremist content, was the inflection point that accelerated investment in brand safety technology and made keyword blocking a standard part of campaign setup.
The technology that emerged — primarily from IAS, DoubleVerify, and Integral Ad Science — uses keyword-based classification to flag pages for advertiser review or to block advertising outright. A page containing words from a block list — typically words associated with violence, drugs, illegal activity, political controversy, or explicit content — is classified as unsuitable for advertising and ads are suppressed.
This approach has real value in the contexts it was designed for. Preventing ads from running adjacent to terrorist recruitment videos or graphic crime reporting is a legitimate advertiser interest and a technically achievable goal. The problem is that keyword blocking is not contextual analysis. It is string matching. It finds a word and applies a rule. It has no capacity to determine whether the word appears in a thoughtful health policy article written by a Pulitzer Prize-winning journalist or in a piece of inflammatory social media content.
The Scale of the Current Damage
Since mid-March, a substantial majority of the most-read pages on the internet contain keywords that are on most advertisers’ COVID-related block lists: coronavirus, COVID-19, pandemic, outbreak, quarantine, lockdown, and dozens of variations. The Reuters Institute at Oxford has documented the scale of the news consumption surge — UK news consumption is up by roughly 60 percent, with similar patterns across major markets.
According to reports from multiple publishers and programmatic revenue platforms, COVID-related pages — which at many news organizations now represent the majority of their total page inventory — are showing fill rates and CPMs 50 to 70 percent below their normal levels. Some publishers report that certain coronavirus-specific landing pages are seeing CPMs below $0.50 in open exchange environments, compared to $3 to $5 for comparable news inventory in normal conditions.
The Guardian, Financial Times, BBC, and dozens of other publishers have spoken publicly about the impact. Local and regional news organizations — which have less diversified revenue and fewer direct advertiser relationships to fall back on — are in some cases facing existential financial pressure from the combination of COVID advertiser pullbacks and keyword blocking.
Why the Overcorrection Happens
The advertiser behavior driving this is rational from a narrow risk-management perspective. Brand managers and media buyers who approve keyword block lists are responding to a real concern: no major consumer brand wants screenshots of its ads appearing next to death toll graphics or overwhelmed ICU coverage in social media posts criticizing advertiser callousness. The reputational risk is asymmetric — no one gets fired for being too careful, but a single screenshot can generate a news cycle.
The problem is that “too careful” has become structural and self-reinforcing. Block list vendors provide default lists that are conservative by design. Agencies apply those lists as baseline campaign settings. Advertisers approve them without detailed review. The result is that the intersection of “contains any COVID keyword” and “acceptable advertising environment” has collapsed, even though most COVID-related editorial content is exactly the type of high-quality journalism that brand safety was ostensibly designed to protect — not prevent.
IAS and DoubleVerify have both issued guidance in recent weeks urging advertisers to move toward more nuanced classification. DoubleVerify published a statement encouraging brands to work with their teams to review block list scope and consider contextual classification rather than keyword presence as the trigger for blocking. The sentiment is correct. The practical adoption is lagging.
The Case for Contextual Maturity
The solution to this problem is not to abandon brand safety. It is to graduate from keyword-based block lists to contextual analysis — understanding what a page is actually about rather than whether it contains specific words.
Contextual targeting technology has advanced significantly in the past three years. Vendors including Grapeshot (now Oracle), Peer39, and GumGum’s VERITY platform use natural language processing and semantic analysis to classify pages at the topic level, not the keyword level. These systems can distinguish between a page about COVID public health policy and a page about COVID mortality rates and ventilator shortages — and make different advertising suitability determinations for each, based on advertiser-defined brand values rather than blanket keyword rules.
The IAB’s LEAN principles — Light, Encrypted, Ad choice supported, Non-invasive — articulated a vision for advertising that respects publisher environments and audience experience. The brand suitability equivalent of that vision is an advertising industry that can make nuanced judgments about content context rather than defaulting to word-matching when editorial complexity increases.
This is not a new argument. The programmatic industry has been having this conversation since 2017. But COVID is making the cost of not having resolved it impossible to ignore. The journalism that is being demonetized right now is not toxic content. It is the most consequential public interest reporting in a generation.
LEAN Principles Revisited
The LEAN framework also has something to say about the advertiser relationship to publisher environments. The fourth principle — Non-invasive ads that don’t interfere with user experience — implies an advertiser responsibility not just to the user but to the publisher ecosystem that connects advertisers to audiences. An industry that systematically destroys the revenue base of the publishers it depends on for reach is engaged in a slow self-demolition.
The argument for publishers to diversify away from open exchange programmatic revenue is being made urgently and correctly by a number of media industry analysts right now. But the argument for advertisers to invest in contextual maturity and review their brand safety protocols is equally urgent — not out of charity to journalism, but out of commercial self-interest. The long-term health of the open web’s publishing ecosystem is a condition of the long-term health of programmatic advertising as a buying channel.
For buyers and agencies reading this: the most concrete action available today is a keyword block list audit. Pull your current lists. Identify every COVID-related keyword that is blocking news publisher inventory broadly. Ask your brand safety vendor what contextual classification alternative is available. The answer exists — the question is whether there is organizational will to implement it.
FAQ
What is the difference between brand safety and brand suitability? Brand safety is the practice of preventing ads from appearing adjacent to content that is clearly harmful to a brand — extremist content, graphic violence, illegal activity. Brand suitability is a more nuanced concept that accounts for a brand’s specific values, voice, and audience and evaluates whether a given piece of content is appropriate for that brand in particular, even if it is not generically unsafe. A brand that focuses on health and wellness might consider appearing adjacent to health crisis news to be brand-suitable even though a luxury fashion brand might not. Brand suitability requires more sophisticated classification technology and more intentional brand configuration than simple keyword blocking.
Are any DSPs offering contextual brand suitability tools right now? Several DSPs integrate with contextual classification vendors — Oracle’s Grapeshot, Peer39 (now owned by Publicis’s Epsilon), and IAS/DV contextual segments — that can provide topic-level rather than keyword-level classification. Whether your DSP has this available as a blocking or targeting option depends on the platform and your current contract. Ask your platform team specifically about contextual page classification options as an alternative to keyword block lists.
Should news publishers be using private marketplaces to protect against this? Yes, and many already are. Publishers with strong direct advertiser relationships who have set up private marketplaces with floor price commitments are insulated from the open exchange CPM collapse to the extent those deals honor committed spend. The challenge is that many news publishers — especially local and regional organizations — do not have the sales infrastructure to build meaningful PMP businesses and are heavily dependent on open exchange programmatic revenue. For those publishers, the short-term answer is advocacy with brand safety vendors and industry bodies; the long-term answer is audience-driven subscription business models.
Is there a way to support quality journalism through this period programmatically? Some buyers have added “inclusion lists” — curated lists of premium publisher domains — to their campaigns specifically to ensure that budget is flowing to quality journalism environments during the pandemic. This approach, which bypasses keyword blocking by specifying trusted publisher domains rather than relying on content classification, is more cumbersome than contextual classification but is available in most DSPs. The Trusted News Initiative publisher list provides a starting point for inclusion list curation.