Reach and IBM launch new brand safety platform for advertisers
Reach, owner of The Mirror and one of the UK's largest commercial publishers, has teamed up with global computing giant IBM to launch Mantis, a new brand safety platform.
The new product is underpinned by IBM Watson Natural Language Processing and Watson Visual Recognition on the IBM Cloud, and uses AI and machine learning to check whether content is appropriate for advertising and reducing unnecessary blacklisting.
Today a significant proportion of news content alone is blacklisted to advertisers, reducing revenue for publishers and reducing reach within premium content for brands, a problem that stems from the less intuitive and less sophisticated solutions currently on the market.
By sitting on the IBM cloud, Mantis will enable Reach to develop secure brand safety solutions at scale supporting IBM and Reach’s goal of rolling out across the globe.
Mantis comes at a time when brand safety has been an increasing concern for the advertising industry, with publishers and social platforms often unable to monitor content at the volume and accuracy required to guarantee a piece of content is safe for a brand to be placed within.
Brands and agencies have also expressed frustration with the need to increasingly bring in third party brand safety tools themselves.
Existing models for identifying unsafe content have largely relied on the creation of long lists of “unsafe” keywords.
However, the model has been far from bulletproof and has resulted in the blacklisting of reams of undoubtedly safe content, resulting in lost ad revenue for publishers.
Terry Hornsby, Digital Solutions Director at Reach, believes the solution lies with the publishers themselves, and needs to be much less obstructive and much smarter and more sensitive to context.
“From a brand perspective, the margin for error has simply been too great – no one wants to see their product next to an upsetting or graphic story”, Hornsby said.
And from the publishing side, simply too much of our content was being blacklisted, even the most perfectly innocent stories.
“My starting point was making sure that a football report on someone ‘shooting a winning goal’ was recognised for what it is – a great piece of content that most brands would love to be next to, not a violent story.”
Reach initially partnered with the IBM Garage to develop the concept, using user-centric Enterprise Design thinking, and agile principles, and, collaborated on a business case to support the development of a minimum viable product (MVP).
The MVP was rapidly integrated into Reach’s day-to-day operations and workflows, by continually being trialled and refined with a feedback loop, using thousands of real-life examples and with the expertise of Reach’s journalists and media experts.
The development team used IBM Watson Natural Language Processing and Watson Visual Recognition to build a model that could automatically read and understand the context of articles and images.
The combined team created machine learning models based on industry-standard taxonomies of blacklisted URLs and keywords, along with its own proprietary catalogue of techniques for identifying ad-friendly content.
Critically, the AI-powered solution also provides the rationale for its decisions, helping human reviewers tweak and improve the decision-making process as it develops and learns and gains even greater accuracy over time.
In a practical sense, this means using image recognition to instantly flag the context of a piece and much more sophisticated language recognition to capture nuance and context.
Using these tools, a “nude lipstick” is no longer marked as salacious and a story about a new life-saving “drug” is no longer flagged as criminal.
Accurate detection of content
The solution has allowed for much more accurate, and automated detection of whether the content on a page is ‘brand safe’ and even at high volumes, without any publishing delay.
As a secondary use, Mantis can also be used as a tool for editorial teams to see instantly if their content may be unfairly deemed unsafe because of certain words, ultimately putting the power in their hands.
The platform is currently being integrated with Reach’s CMS across the editorial teams for this purpose.
Alison Davis, IBM Industry Lead for Telco, Media and Entertainment in the UK said: “Safeguarding an organisation’s brand is essential to earning consumer trust and publishers globally have struggled to use the existing solutions to prove that their content is safe for those brands.
“By bringing this digital platform to market, IBM and Reach aim to improve the advertising landscape for both publishers and brands.
“It’s only by deploying AI with IBM Watson, that we can help Reach detect brand safety issues in real time and help prove how not only safe, but effective content that’s blacklisted today, can be for advertisers.
“It’s a win win which we believe will scale across the industry in the UK and globally.”
The product is now live at Reach and is being taken out to market for other publishers and content owners, both in the UK and abroad.
Reach is handling licensing queries in the UK while IBM is covering the US market.