In our recent ‘Protect your brand’ webinar we received a lot of great questions about the different aspects of brand safety. (If you missed it, you can access the recording here.) In this blog post, we’re answering your brand safety questions.
What is one aspect of brand safety that you don’t think advertisers are thinking about and that they should be?
Different brands have different topic areas that they don’t want their advertising to serve next to, for example, Disney would not want to serve next to an article about a new nightclub, whereas an alcoholic drinks brand won’t mind serving next to the same article. All brands have different preferences, and what each brand is trying to achieve will be slightly different. Some would like lots of different brand safety models to address their specific needs. For example, one of the rising areas that we have seen is children’s content and that is where we may likely see the focus shift to next. For example it is not appropriate for alcohol brands to serve next to children’s content, whereas toy brands may benefit from serving next to content targeted at children. Advertisers should think about their tolerance to risk and work with IAS to ensure their brand reputation is fully protected online.
How is keyword blocking used in regards to risky online content?
Firstly, it is good to understand that keyword blocking can be used in a few different ways. For IAS keyword blocking, we look at specific terms in the URL that advertisers may be sensitive to. Another technique that we use to protect advertisers it to look for keywords in new articles, these keywords are often included in the URL for search purposes and help set a keyword list. We then put parameters in place that say if ‘X’ keyword appears in a URL then don’t serve an ad. Keyword blocking on the page is an element of our data science-led brand safety approach, we believe that keywords are one piece of evidence on a page, or in content, but that they are not the only piece of evidence used.
To ensure we have the most effective brand safety solution we always want to understand if clients identify a risk to their brand that’s not being accounted for in our seven brand safety content models. If we see that is the case then we work with advertisers to address this and ensure they are protected. We then take these insights as learnings, and incorporate them into our overall data science-led approach. Updating our brand safety models rather than building static lists of keywords and looking for those on a page, then rate the page poorly. This approach can be effective but also short-sighted, because in some cases the page may actually be brand safe but the keyword is used in another context. It can unnecessarily restrict advertisers’ scale.
How do your systems localise to content in different languages and factor in cultural reference?
Our brand safety model works across over 40 languages, from Arabic to Vietnamese, and encapsulates seven different risky content categories, such as adult content or illegal downloads. Words that would be blocked in English such as, terrorist, and extremist can now be blocked in the most commonly found languages on the internet, and cover 99% of content found online. As new slang words arise in local markets, and as languages evolve, we work diligently with local market linguist experts to update words in our models.
Do you see more interest in brand safety for Brand campaigns. Or do you see brands with strong DR focus have a different approach?
We do not see that there are major differences in the approach for DR or brand campaigns. Brand safety is subjective, what constitutes as high risk for one brand may be seen as low risk for another. In brand campaigns, we do see many differing requests depending on the brand’s preference and tolerance to risk. Take a famous beer brand as an example, they would not want to block against alcohol content but a family brand such as baby care goods would block against such content. Every brand will have a different set of keywords that they want to block against and every brand will have different websites on their exclusion and inclusion lists.