GumGum’s latest analysis reveals how contextual advertising tools reshape political campaigns by uncovering significant media trends and sentiment shifts, offering strategic insights for tailoring messaging and targeting.
This recent political season has been full of twists and turns, and that doesn’t seem to be changing anytime soon. We’re all looking for ways to contextualize this presidential election cycle to keep our heads wrapped around what’s happening.
A recent analysis by GumGum sheds light on how the media portrays Vice President Kamala Harris and Former President Donald Trump in political discourse. Utilizing their advanced contextual advertising tool, GumGum examined over 5 million pages of political content from late July to uncover trends in media coverage and sentiment.
The findings reveal a significant disparity between mentions and sentiment: the media mentioned Donald Trump more than 1.7 million times, while Kamala Harris was mentioned around 1.36 million times. Notably, Kamala Harris’s media presence surged by 388% on the day of her candidacy announcement.
We chatted with Hailey Denenberg, VP of Strategic Initiatives, Data at GumGum, to explore these findings and how contextual advertising can help political advertising campaigns.
Leveraging Contextual Analysis for Political Campaigns
Andrew Byrd: Why did GumGum decide to conduct this study? What were your goals, especially in relation to your contextual tool?
Hailey Denenberg: Our contextual technology, developed over the past 15-16 years, uses computer vision and natural language processing to understand content as a human would. With the upcoming political campaign, we wanted to leverage this technology to track trends over time.
Given our extensive programmatic integrations with large platforms like DSPs and SSPs, we have a vast amount of classified content at our disposal. Following Kamala’s presidential announcement, we found it particularly interesting to analyze how the open web and editorial content discussed Kamala versus Trump over the last two weeks in July. We focused on mentions and sentiment, aiming to understand how each candidate is perceived and discussed in content.
Understanding how different editorial voices and platforms perceive and talk about each candidate, especially in terms of sentiment, provides valuable insights into the broader public discourse. Our goal was to use this analysis to uncover trends and patterns in how content about these candidates is produced and consumed, ultimately offering a deeper understanding of the political environment through the lens of digital content.
AB: The upcoming election is full of unexpected developments, especially after Kamala’s recent announcement and the campaign’s strategic execution. Given this dynamic, can you explain how contextual analysis helps understand political coverage?
HD: Contextual analysis has many applications, especially in presidential campaigns. Advertisers should stay updated on how content is trending, positively or negatively. Think of it as playing offense and defense. For instance, if mentions of Kamala’s opponent, Trump, spike negatively, her campaign can play offense by surrounding that content with positive messaging about Kamala.
On the other hand, they might want to avoid any negative mentions of Kamala to keep the messaging streamlined, which is more of a defensive strategy. Understanding these content trends as different announcements unfold allows campaigns to use advanced contextual targeting. They can target all positive or negative content that mentions specific candidates, leveraging this analysis to shape their messaging effectively.
AB: How can publishers benefit from these technologies?
HD: Yes, publishers can significantly benefit from contextual technologies. For example, news publishers can strategically package their inventory by grouping positive political news and offering it to brands comfortable with political content but wanting to avoid association with sensitive issues like abortion or immigration. This allows them to monetize content that aligns with the advertiser’s brand safety requirements.
Advanced contextual technologies not only understand the sentiment of the content but can also identify and filter out specific sensitive topics that an advertiser may want to avoid. This capability is crucial for maintaining brand safety while still allowing advertisers to participate in positive, relevant conversations.
Addressing Brand Safety Concerns
AB: At our last conference, there was a significant discussion about brand safety. Advertisers naturally want to avoid being associated with certain content, but publishers face revenue challenges due to these restrictions. How does GumGum approach brand safety, especially in contextual advertising?
HD: Brand safety has become even more critical recently, especially with the news around the dissolution of VM and GARM. While we align with the GARM framework, we’ve also developed our custom threat categories beyond the usual “Dirty Dozen” like violence.
This is important because while positive contextual targeting aligns with preferred categories, there’s a strong demand for blocking or negatively targeting specific categories. Where GumGum stands out is in video analysis. For instance, in political advertising, which is huge for TV and CTV, most providers struggle to analyze what’s happening within the video content.
However, our advanced video contextual technology allows us to explore the complete audio transcription and perform scene-by-scene or frame-by-frame analysis. This helps us determine what parts of the video are brand-safe or suitable according to our established threat levels, ensuring more precise targeting and reducing over-blocking.
Reaching the Right Audience and Environment
AB: How does contextual advertising help political ads reach the right audience in the right environment?
HD: At GumGum, we focus on placing ads where they are most relevant, using a deep understanding of content. For example, if someone is reading an article about the election and sees an ad with positive messaging about a candidate like Trump, it aligns with what they’re already thinking about.
This increases the ad’s effectiveness, especially when combined with geo-targeting in swing states. It’s all about reaching consumers at the moment they’re considering a topic, which can influence their actions.
AB: Does your approach to contextual advertising change depending on whether it’s on mobile, desktop, or other platforms? Or is there generally much overlap?
HD: Generally, there’s significant overlap, as reaching the consumer in the right mindset is often based on the content itself, which tends to be consistent across devices like mobile and desktop. However, attention models can vary depending on the environment. We have different panels for mobile and desktop, and we’re beginning to explore video. While I don’t have definitive data yet, it will be interesting to see if there are differences in optimal attention times across these environments.
AB: What final advice would you give to advertisers and publishers considering political advertising and contextual targeting?
HD: I advise them to gather as much data and insights as possible from various sources. This will help you uncover unconventional ways to position your brand or supply. Instead of sticking to obvious choices, explore new audience segments. For instance, while Nike might seem like a fit for sports content, insights might reveal it’s also popular in Home and Garden content due to a current trend. By broadening your perspective, you can enhance your targeting strategy and reach new audiences more effectively.