Brand Safety
Keeping ads away from unsafe content with categories, keywords and sentiment.

Brand Safety Solutions on Eskimi
Brand safety refers to the measures taken to ensure that a brand's advertising will not appear alongside content that could harm the brand's reputation or be offensive to its target audience. This can include content that is violent, hateful, sexually explicit, or promotes illegal activities. 

 Brand safety is an important consideration when placing advertising, as it helps to protect the brand image and reputation. It can also help to prevent legal issues from arising from advertising being placed next to inappropriate content. This can be achieved through various ways such as using brand safety tools, monitoring the placements of ad, using keywords filtering, context analysis and more. 

 Eskimi DSP supports several layers of brand safety tools which ensure a relevant and brand-safe environment.  

 

 Traffic source content information 

 Category exclusion 

 The first layer consists of checking the general website content from the information provided by the exchanges and/or publishers. All placements are categorized, and sensitive categories - such as adult content, betting, arms, violence, etc. - are excluded accordingly in every campaign:  

 Picture 1 - Categories that excluded in Eskimi DSP as default 

 Pre-optimization based on historical data 

 Besides the category exclusion, campaigns are pre-optimized by using historical placement stats to exclude sites or/and apps that showed poor performance in the past. Also, this tool works both ways – it is possible to run on sites/apps that showed good performance. In other words, general brand safety solutions are done for each brand and its advertising. 

 Limitations:  Information that comes from an exchange/publisher is not always accurate. 

 Addressing limitations:  Employing additional internal tools to ensure brand safety. 

 

 URL Keyword checking 

 Unfortunately, traffic source content information isn’t enough when facing placement content. A lot of news sites are categorized as not sensitive, although they are full of sensitive and disturbing information that may harm the brand. To move forward with an in-depth solution, Eskimi DSP launched a URL keyword checking tool. If an URL contains keywords from a negative keywords list, an impression will not be purchased and the ad won‘t be showed near such an article. 

 Limitations:  1) URL doesn’t include any negative keywords. 2) Apps traffic cannot be checked. 

 Addressing limitations:  1) Reviewing website content. 2) Whitelist/Blacklist for apps. 

 

 Site content reviewing for keywords 

 The keyword checking an URL helps to analyse the content. However, it doesn’t safeguard from lists that don’t have a negatively listed keyword in the URL but have a lot of unwanted material in the content itself. To bring brands to an even higher brand safety solution, Eskimi DSP also analyses page content: keyword lists set to "Find in content" check the text of the page, not only its URL, and our AI assigns categories to individual pages rather than whole domains. 

 Limitations:  1) Content doesn’t include any negative keywords. 2) Apps traffic cannot be checked. 

 Addressing limitations:  Whitelists/Blacklists 

 
 Sentiment targeting 
 Eskimi's AI reads page content and labels its overall tone as Positive, Neutral or Negative; inventory that hasn't been analysed yet is labelled Unknown. Content can be in a safe category and still be negative in tone, for example a news article about an accident or a crime. Sentiment targeting lets you keep ads away from that content by running only on Positive, Neutral and Unknown inventory. Unlike keyword lists, sentiment also covers app traffic. See How to Enable Brand Safety on a Campaign Level for the setup. 
 Limitations: 1) Leaving out Unknown removes inventory that hasn't been analysed yet, which can noticeably reduce reach. 2) Tone is not the same as safety: a negative article can be fine for some brands, and a neutral one can still be unsuitable. 
 Addressing limitations: Combine sentiment targeting with category exclusion and keyword lists. 
 

 Human curated whitelist/blacklist 

 By excluding sensitive topics, we limit ads to certain categories. However, some brands are solely focused on exact categories. On Eskimi DSP it is possible to whitelist and blacklist sites/apps based on brand preference. This means that ads will appear or won’t appear on (un)wanted placements. 

 Limitations:  1) Different definitions when understanding avoidable content. 2) Whitelist limits traffic. 3) Human mistakes. 

 Addressing limitations:  Align and approve whitelist with the client. 

 

 External tools & services 

 In addition to our in-house brand safety solutions, we are working with 3 rd  party brand safety platforms such as DoubleVerify, Integral Ad Science, The Media Trust, and many other tools if a client requests. 

 When DoubleVerify is enabled for the advertiser, DoubleVerify pre-bid inventory segments can also be selected directly in the campaign, in the DoubleVerify Inventory Segments section, so unsuitable inventory is filtered before the bid. 

 To sum it up, brand safety should be one of the highest priorities for the brand and for the platforms that they are partnering with. With an always growing demand, Eskimi DSP is working on in-depth solutions to fill the client’s needs. As the market is getting more aggressive and competitive these deeper solutions are key when delivering good services. 

 Limitations: 1) Extra costs when using selected tools. 2) Potential increase in CPM price. 3) Limited abilities to verify app traffic.

How to Enable Brand Safety on a Campaign Level
Brand safety refers to the measures taken to ensure that a brand's advertising will not appear alongside content that could harm its reputation or be offensive to its target audience. This can include content that is violent, hateful, sexually explicit, or promotes illegal activities. 

 

 Brand Safety for your campaigns 

 You can ensure Brand Safety for your campaigns by using keyword lists, apps/sites categories and sentiment targeting, separately or together. These features work together to safeguard your brand's image and prevent your ads from appearing near undesirable content. 

 How to exclude Keywords for your campaigns  

 To enable brand safety feature in your campaign, you need to: 

 

 

 Create a keyword list of certain words you want to exclude. 

 

 

 Enable Brand Safety on a campaign level. 

 

 

 Keyword Lists 

 The keyword list can be created on the Keywords page: 

 

 

 Visit the Keywords page under Tools. 

 

 

 Click Create at the top right. 

 

 

 Add “Title” to your Keyword list. 

 

 

 Select at least one “Context option”. 

 

 

 If you choose “find in URL”, the system will check if the keyword is in full URL and exclude the URL. E.g. if you exclude a set that contains murder, and a bid request comes from randomsite.com/murder, Eskimi will not bid on this request. 

 

 

 If you choose “Find in content”, the system will crawl relevant campaign sites and only show ads if the domain was crawled and no ‘murder’ was in that domain. E.g. if randomsite.com comes in a bid, Eskimi will not bid till our crawler checks what’s in that site. Once we do the check, we will see if ‘murder’ is part of the content or not, and include/exclude the site accordingly. Note: both of these options can be selected. When both options are selected, the URL and page content are checked for keywords. 

 

 

 

 

 Select “Match option”. 

 

 

 Contains type match: looks for similar looking words. if you have ‘murder’ in the set, all similar words will be added: ‘ murder ers,  murder ous, randomsomething murder , etc.’. 

 

 

 Exact word type match: looks for exact words. If you have ‘murder’ in the set, it will only look for ‘murder’, not ‘murderers’ or ‘murderous’. 

 

 

 

 

 Upload keywords. 

 

 

 Click Save. 

 

 

 Check if your keyword set doesn’t have single letters, words of a few letters or similar. Such setup would cause “contains type match” to exclude lots and lots of traffic. E.g. if you added letter ‘a’ to the set and chose 'contains type match', all sites containing letter ‘a’ would be excluded. That can be a lot of sites. 

 Keywords only work on full URLs. e.g. randomsite.com/randomsite, randomsite.com/notsorandom, randomsite.com are treated as 3 different sites. 

 Excluding keywords on a Campaign 

 Once you have created a set of keywords, you can enable brand safety for your campaign. 

 

 

 Open the selected campaign. 

 

 

 Scroll down to Brand safety. 

 

 

 Select one or more keyword sets under the Keywords targeting. 

 

 

 Click Save. 

 

 

 

 Keep in mind that if contextual targeting  is enabled on the same campaign, both settings will be taken into account. If both brand safety and contextual targeting keyword sets have the same words, the matching keywords will be excluded. Eskimi ensures that the Brand Safety setting is always prioritized. 

 Once the brand safety settings are set on a campaign: - If the keyword set is 'find in URL', the campaign starts using the URL for validation. - If the keyword set is 'find in content', the crawler kicks in and starts checking relevant sites, and bidding only once relevant sites are checked. So there might be a delay in checking what content is hidden behind the site URL. Please note that this feature is only available for site placements and not applicable for apps. 

 How to exclude negative sentiment content for your campaigns 
 Content with a negative tone, such as reports of violence, disasters or crime, can sit in categories that are not marked sensitive. Sentiment targeting keeps your ads away from it. 
 Sentiment targeting works by selecting the sentiments the campaign may run on. To exclude negative content: 
 
 Open the selected campaign. 
 Scroll down to Contextual targeting . 
 Under Target sentiments , select Positive , Neutral and Unknown . Leave Negative unselected. 
 Click Save . 
 
 Keep Unknown selected unless the brand requires it: it covers inventory that hasn't been analysed yet, a noticeable share of impressions, so leaving it out reduces reach. Sentiment is checked on both sites and apps. More on the sentiment options: How to enable Contextual Targeting on a campaign level . 

 How to exclude apps/sites categories for your campaigns  

 The apps/sites category feature helps users target ads based on the market verticals of the page. Apps/sites categories contain already pre-created lists of domains and app bundles of related topics such as Finance, Sports or Games. Categories are grouped into two bigger groups - General topics & Sensitive topics accordingly.  

 To setup category targeting, you need to: 

 

 

 Open the selected campaign. 

 

 

 Scroll down to Brand Safety. 

 

 

 Select categories under the Apps/Sites categories. 

 

 

 Click Save. 

 

 

 

 Keep in mind that if apps/sites categories targeting is enabled under contextual targeting on the same campaign, both settings will be taken into account. If both brand safety and contextual targeting have the same categories selected, the matching domains or apps bundles will be excluded. Eskimi ensures that the Brand Safety setting is always prioritized. 

 Our categorization system now works on two levels: - Page-level: Our AI engine analyzes individual page content to determine precise categories (e.g., a sports article on a news site will be categorized as "Sports") - Domain/app bundle level: Used as a fallback when page-level analysis is not yet available The system prioritizes page-level categorization for better targeting precision, while maintaining domain-level categorization for comprehensive coverage. Note that we continue to crawl page content for keyword targeting as well.