In the aftermath of the 2018 brand safety scandal around toxic adjacencies, many brands discovered they were ill-equipped to fully protect themselves from potential future risks. With the recent announcement from Google that YouTube might never be 100% brand-safe, the onus is clearly on brands to ensure their own brand safety. GumGum's Ed Preedy explores the techniques to do so.
Blacklists and whitelists have proven very popular tools for guaranteeing that advertisements don’t appear where brands know there are risks, with blacklists alone being used by over 60% of marketers. While there is no doubt that these approaches do work, they can also act as a barrier between brands and receptive audiences, reducing reach and ad-spend ROI. Thankfully, the wide range of tools available that are dedicated to ensuring brand safety, means that it is possible to ensure brand safety without limiting your audience.
The focus for brands should be on educating themselves about which tools will work best for them. Brand safety is a layered issue, with threat vectors varying widely across industry, audiences and geographies, but below are a few of the more powerful tools that can help brands keep themselves secure.
Direct relationships with publishers
Buying ad space directly from a reputable publisher allows marketers to be secure in their brand safety from the get-go. Having a direct line to someone you can trust at a publisher also allows for quick changes to made on the fly when concerns do arise, making it possible to react to emerging threats before they escalate. The trade-off with this tactic is the premium price tag that comes with premium placements, and the risk of exposure to hard news topics means remaining vigilant is a must.
Image Recognition (Computer Vision)
An AI-powered technology that uses neural networks to identify and sort images, image recognition is used by 21% of marketers to prevent brand-safety episodes. Given the proper inputs, it can identify an image that constitutes brand unsafe content with an incredible degree of accuracy and automatically keep advertisers out of unsafe environments, or just advise against it if the key decision makers want to take a more hands-on approach. With images being such a prominent medium across the internet, being able to identify those that could be a threat at scale has become increasingly important.
Keyword detection allows for brands to control where not to appear based on the words that appear on a given webpage. This is a valuable tool for avoiding basic types of negative adjacency and are almost a requirement for content rich environments like news sites. Advertisers can choose from hundreds or even thousands of key words to avoid, although expanding an exemption list to this scale will inevitably begin to affect ROI in the same way that a full blacklist will.
Natural language context detection
Often used in conjunction with image recognition to provide an overlapping defence, language context detection software is an evolution of keyword filters used by around 21% of marketers to bolster brand safety. These systems identify brand unsafe text based on the context in which it is used in, rather than just banning based on a word at face value. This ensures that innocent uses of a word are not punished unjustly. It is also useful for filtering out instances where offensive words are presented in a more borderline acceptable way. For example, if a political figure were to use offensive language in an official speech, most news outlets would choose to report the statement uncensored.
While more commonly used for making sure campaigns are running in the right area, brands need to be aware of the real-world negative adjacencies that can occur. In Out-of-home, advertising life insurance outside of a funeral home is not likely to leave an audience with a positive experience, and the same is true in digital advertising. To use a generalised example, if a brand is not well loved in a locality, geo-targeting is a vital tool to be able to ring fence off this location, making sure resources are not spent trying to engage with an audience that could be actively hostile.
Ads.txt (Authorized Digital Sellers)
Definitely one of the more ‘high tech’ solutions on this list, Ads.txt is a security system introduced by the IAB (Interactive Advertising Bureau) specifically for improving transparency around programmatic advertising. An ads.txt file placed in the root of a publisher’s website lists all of the third-party vendors authorised to sell advertising space on a given website. This provides publishers with a means to tell the world who is authorized to sell their ad inventory. While a lot of large ad firms have thrown their support behind ads.txt as a tool for increasing transparency and reducing fraud, it is only as useful as its adoption rate. The more companies that use it, the more effective it becomes.
Many of these tools work best in conjunction with one another, in order to create an overlapping brand safety defence. Image recognition can find brand unsafe content that a keyword identifier would miss, and vice-versa. As the knee-jerk reaction to the brand safety crisis wears off, the trade-off between safety and ROI that reliance on blacklists and whitelists cause will begin to be seen as the false binary choice that it is. The advertisers who succeed in using the right tools to avoid brand safety crises, without sacrificing audience reach, will be the ones to thrive in the long-term.