Feeling Sentimental? AI Understands.

Executive Summary  

From the genesis of social media, companies have incorporated artificial intelligence into their product functionality. From personalized content recommendations to ad targeting optimizations, AI has been continually woven through the fabric of product development. With recent advancements in social media and AI, there has been an emergence of new tools for social teams to utilize. One that has caught the market’s attention in particular is sentiment analysis. This POV takes a look at the historical approach of sentiment analysis, the impact of OpenAI advancement, and CMI Media Group’s recent success with AI tools on paid social campaigns.  

The historic approach  

Before the advent of AI-based sentiment analysis, trends were identified manually by synthesizing data in aggregate while evaluating various metrics. This took hours of work as analysts looked at users’ posts and weighed the negative and positive terms to formulate data points. This manual process was needed due to the limited context that any automation of sentiment analysis could provide. This laborious effort evolved in 2010 with early machine learning iterations of AI detecting forms of expression as “positive” or “negative.” For example, if someone wrote a review with three sentences about their dislike of Brand A but had one sentence at the end about how much they loved Brand B, the sentiment of the review would be marked as “negative.” This initial automation was problematic in the sense that if the analysis was being conducted for Brand B (a positive review) the negative sentiment conclusion was misleading due to the limited advancement of machine learning. The inability to look at the full context is why sentiment analysis remained a mostly manual process up until the release of GPT 3.5 in 2022. 

Sentiment analysis and AI 

The initial integrations of the Generative Pre-training Transformer (GPT) in 2014 allowed language processing to expand from positive and negative sentiments to an emotional dimension identification process. Posts were now categorized by their prevailing sentiment (e.g. joy or fear was expressed). While this was helpful in conducting new forms of analysis it still suffered from a lack of context.  

Fast forward to the present day, industry investment and newer iterations of GPT (3.5 and above) allow systems to uncover the full context of sentiment analysis. Many social listening tools have incorporated these advances by integrating machine learning classifiers to render a deeper understanding of human expression. This means that sentiment analysis tools now have the capability to understand that words can have different meanings from the context that is shared. For example, AI is now able to identify that words like “sick” can describe the state of health or can be used as slang term to describe an extreme sport.  

While adding classifiers is helpful it does require a lot of work to train the data by the end user and is still not completely accurate. To mitigate this, companies are now starting to create integrations with OpenAI to directly connect with their social platform and tools. This has not only increased the accuracy of the sentiment analysis but also has allowed marketers to prompt the AI on the data that teams of analysts have curated, resulting in a substantial decrease in time to create meaningful insights. The caveat still remains that these integrations are not perfect, but there has been a steady increase in accuracy and utility of these tools that looks to continue. 

Social Media and AI  

As brands consider ways to advance their AI social media tool kit, three suppliers to pay attention to are NetBase Quid, Synthesio, and Brandwatch. NetBase Quid’s AI-powered natural language processing (NLP), analyzes real-time social conversations that powers brands with consumer data-driven insights. Beyond spotting trends and viral influencers, Synthesio supports brands with crisis identification and appropriate response generation. Brandwatch evolves the historical approach of sentiment analysis through deep learning by tapping into brands’ portfolio of social networks for the mass consumer opinion. Ultimately, these solutions offer social teams a strong framework of how to reach their target audience in the most effective way.  

Within the greater social sphere, Meta is also testing generative AI elements through the gradual release of AI Sandbox. The new toolbox allows advertisers to simplify tasks like text variation, background generations, and image outcropping to adjust and test different versions of creative and produce more engaging ads with maximized performance. As part of the changes, Meta Advantage, featured in the Business Manager, will also be available for advertisers to leverage AI to create campaigns faster and smarter. This includes switching manual campaigns to Advantage+ shopping, using video creative in catalog ads, and improving performance with Advantage+ audience. For marketers, these add-ons enhance the current campaign tools with the intent of adding more value and efficiency in the optimization process. 

CMI Media Group and AI Advancement  

AI has also transformed the way Twitter/X analyzes sentiments expressed in tweets to better inform audience insights and targeting capabilities. CMI Media Group has seen great success in leveraging X’s AI tools to both understand and reach the HCP audience on paid social. Through the utilization of keyword analysis, CMI Media Group is able to accelerate the understanding of HCP expression and increase campaign targeting precision in real-time by natural language processing techniques. One recent example of this was utilizing a keyword segment of HCPs’ reading and research behavior for polycystic kidney disease to increase audience accuracy and reach. This campaign led to the most cost-efficient results with a significant 74% reduction in cost per impression. As a result, CMI Media Group has expanded this best practice and leveraged these targeting AI techniques across multiple paid social campaigns on X.  

Outlook  

This is just the beginning of AI advancement within the digital marketing industry. From sentiment analysis to image generation, social networking platforms will continually invest in enhancing their AI systems to expand campaign efficiencies. Brands must stay open to these advancements as this is an opportunity to better understand the consumer opinion and build a deeper connection with your target audience. As the AI phenomena develops, CMI Media Group is committed to safely evaluating and implementing these new tools to drive innovation and successful campaign delivery for brands across social media platforms and beyond. If your brand wants to take advantage of the latest AI advancements on social media, reach out to your CMI Media Group social leads to get the conversation started.