As well as some coronavirus content actually being suitable for brands to advertise against, U.K. newspaper group Reach is seeing signs that ads are performing better.
Reach has been using its tool Mantis, which leverages IBM Watson’s machine-learning, natural-language processing to gauge whether content is appropriate, to recoup previously blocked ad inventory on news stories. Now, the tool has been integrated with Xandr’s supply-side platform, making it more easily available to publishers.
You have read the maximum number of free articles.
This content is available exclusively to Digiday+ members.
During March, 52% of content Reach published across its titles related to coronavirus, this rose to 67% in April. Reach had 42 million unique users in March, according to Comscore. It found that between 65% and 75%, respectively, of its coronavirus-related content was neutral or positive in tone, and so suitable for advertising. After all, certain brands are happy to appear next to stories like sports personality Gary Lineker donating two months’ wages to charity, or details of government fiscal stimulus.
“Brands like betting companies and supermarket chains who are advertising right now are extremely sensitive,” said Ben Pheloung, general manager of Mantis. “We can be transparent in what we are looking for and what we are blocking.”
For over two months, publishers’ frustrations have mounted: Traffic is reaching record highs but can’t be monetized because advertisers are wary of bad ad placement, leading to blanket keyword blocking of coronavirus content. With advertisers canceling campaigns, eyes are even more fixed on the bottom line. Trade body Newsworks estimates news publishers’ collective loss of revenue from unwieldy blunt keyword block lists is £50 million ($63 million) in the U.K. over 12 weeks. Furloughs and staff cuts increase the capacity of remaining teams making it hard for businesses to focus on tackling industry-wide issues.
Lack of clear evidence of the how well ads perform in news environments — despite studies from Comscore and Newsworks and many others — has led to advertisers favoring blunt keyword lists to protect their ads.
“A lot of the brands who have been advertising are trending away because [ads are] blocked, but we do see higher dwell time and click-through rate,” said Pheloung. “People are more engaged with the ads.”
For one campaign Reach ran across its titles comparing coronavirus-safe segment scanned using Mantis with a client’s normal brand safety controls, the Mantis version saw a 26% increase in click-through rate.
On a weekly-running campaign, Reach compared blocking all coronavirus content one week, using that data as a control, and then using the Mantis coronavirus-safe segment the following week. The in-view time for the mobile MPU, which was the most common campaign format, increased by 48%, according to the publisher. The creative message was slightly different each week, but the brand was the same.
While this speaks to the touted network effect of news brands, it’s also because coronavirus content is pulling in lots of readers, readers who may be less used to seeing ads run here.
“We haven’t been able to measure the impact of advertising on coronavirus-related articles versus non-corona articles,” said an executive at a U.K. news publisher. “But we do know that coronavirus articles have on average a 25% longer dwell time, 2.5 times as many comments and nearly 4 times as many shares.”
For now, more than five soon-to-be-announced U.K. and international publishers are using Mantis, although Pheloung wouldn’t say which, as part of a commercial partnership between Mantis and Xandr. Reach has previously told Digiday that it’s not expecting to drive major profits from licensing the tech. While publishers are scrambling to find ways to trim costs, signing new tech licensing deals may not be top of the list.
“This [partnership] allows us to apply Mantis broadly in a way so publishers can adapt quickly,” said Pheloung. The amount of coronavirus content that Reach found was suitable for brands mirrors that of other publishers that have been using the tool, he said.
Phelong said it’s trying to replicate the open marketplace as much as possible. Xandr passes news URLs through the Mantis filter to be scanned, Mantis then returns all of the brand-safe URLs. Then, Xandr aggregates the targetable inventory across multiple news publishers to make it available to buyers as a package.
While custom-publisher solutions are appealing, a lot of publishers need to be on board so agencies needn’t split their brand-safety strategies for different media plans, said Matthew McIntyre, vp of programmatic at Essence agency.
“Even on something like a PMP-basis we would still need to look at how to segment out that part of the campaign and remove standardized keyword blocking across others,” he said. “Keyword blocklists are blunt but they are used because they can scale across whole campaigns. It’s hard to envisage a silver bullet that’s not directly involved with the advertiser. You have to have multiple opportunities to find the right mix, this definitely adds to that.”
Harsh keyword blocking methods are being used in direct insertion order deals, programmatic guaranteed deals, private marketplaces and on the open marketplace. The counter is that low quality, low yield inventory on the open marketplace naturally exposes advertiser to more risk, overt blocking is the symptom of a race to the bottom. As the programmatic open marketplace has dried up, Reach, as well as other publishers, has seen an increase in direct deals as brands want to make sure they have the right brand safety controls in place during coronavirus.
“We’ve passed the initial knee-jerk block everything stage,” said Pheloung. “Brands that are advertising are exploring how they can advertise more effectively and understand a little better the state of the world we live in.”
Sign up to get the day’s top stories at 6am eastern.