Mark your calendar for Mediaweek, October 29-30 in New York City. We’ll unpack the biggest shifts shaping the future of media—from tv to retail media to tech—and how marketers can prep to stay ahead. Register with early-bird rates before sale ends!
U.K. publisher Reach plc, which operates more than 130 news sites including The Mirror and Manchester Evening News, announced Monday the commercial debut of a content recommendation tool. Called Neptune Recommender, the product uses machine learning and first-party cookies to serve visitors targeted suggestions for further reading.
Internally, the tool aims to lift advertising revenue across the Reach plc portfolio by increasing page views and dwell time. Here, it’s already seen gains—the tool accounts for 40% of the media company’s page view growth this year, according to group digital and innovation director Terry Hornsby. The publisher declined to share the financial impact of the product, which was developed internally by a team of five.
Reach plc also plans to make Neptune Recommender available to other publishers, in some cases charging a revenue share from the page views it generates.
The launch comes amid a larger push from publishers to improve the targeting capabilities of their contextual offerings, as the looming deprecation of third-party cookies promises to render obsolete the practice of serving personalized ads based on browsing history.
“Trending has its place in the world, but the web is greater than that,” Hornsby said. “We’ve got to be relevant for people for the long term rather than just in the moment—and that’s what this product has achieved.”
Brand safety tools recommend articles more precisely
Neptune Recommender works in tandem with Mantis, a brand safety tool developed by Reach plc in partnership with IBM Watson in late 2019.
Mantis, which Reach plc also licenses commercially, uses machine learning to scan articles and gather data points such as category, concept, comment engagement and sentiment, allowing it to determine the subject of an article more accurately. By combing through text more granularly, Mantis lets brands distinguish between, for example, a depressing article about Covid-19 versus an uplifting story related to the pandemic.
By harnessing that technology to provide reading suggestions, Neptune Recommender is able to pinpoint recommendations that would be impossible when only using keywords. For instance, if a reader browses an article about minimizing food bills, Neptune Recommender will not only recognize the category of the article––cost of living––but also the concept of budgeting grocery costs.
The product combines those contextual insights with first-party cookies, which refresh with each article consumed but disappear after seven days, according to Hornsby. This way, when someone lands on an article, the product makes recommendations informed by both their current and recent reading history. If a reader logs into their account before browsing, Neptune Recommender also pulls from their profile data.
The widget itself resembles a content carousel, and it can recommend up to 17 related articles. It typically appears after the fifth or seventh paragraph in an article and works on mobile, desktop and Google Amp; the publisher plans to incorporate Neptune Recommender into its 45 apps by the fourth quarter.
Prior to creating the recommendation tool, Reach plc had no automated system for suggesting relevant articles to readers. Instead, it relied on manual recommendations written by editors and other widgets, like trending or top stories.
Offering hyper-targeted contextual ads
Reach plc hopes to see the Neptune Recommender increase its digital advertising revenues in two key ways.
First, the media company will be able to serve highly targeted ads to its readers without using third-party cookies. This could let Reach plc generate more digital ad revenue on browsers that already forbid third-party cookies, such as Mozilla Firefox and Safari—and it positions the company to more seamlessly handle the planned abolition of third-party cookies on Google Chrome next year.
Neptune Recommender also introduces a new ad product to the Reach plc portfolio: targeted ads within the carousel widget. Given that the recommendation tool suggests content based on the current and historical reading history of visitors, the media company can sell the surface area to brands whose products align with these narrowly determined interests. It can also serve different categories of ads within the carousel.
The product allows brands to sponsor specific subjects, such as a regional soccer team, the Eiffel Tower or the rising cost of food, Hornsby said. Buyers can choose from 2,000 categories and 600 concepts, entities and emotional sentiments.
“Brands want to get specific messaging across in a specific type of context,” Hornsby said. “Now you don’t have to sponsor an entire section to achieve that.”
