A Measure of LLMs’ Propensity to Recommend Brands and Products
Influencer marketing exploded with the mainstream adoption of social media platforms by consumers. Brands were quick to embrace consumers’ desire to get product recommendations from people like them, or to embrace brands promoted by people they admire. Social Media platforms have supported influencers to rapidly expand their reach and their scale and influence. Measurements like Net Promoter Score (NPS) and Social Sentiment allow Marketers to track brand health across the social media platforms.
As Gen AI transforms how we search for Answers, LLMs are poised to become the new influencers of consumer purchase decisions. Monitoring brand representation within the major LLMs is an important new skill set within the marketing discipline. We believe that devising a new metric that defines the likelihood of an LLM to recommend a brand on key prompts becomes a critical tool in this new world.
Introducing the Revere Brand Index Score (RBI) – a new metric now available in Brand Luminaire that synthesizes the likelihood of a brand to be mentioned in response to key prompts, and the likelihood that this representation will be favorable. As an example, below is a table of Brand Index Comparisons for the Electric SUV market.
RBI allows marketers to quickly gauge Rank (likelihood their product brand is included in the recommendation set) as well as overall Favourability sentiment in the LLM answer. The sum of these two metrics equals the Revere Brand Index score. In short, the RBI helps brands understand the propensity of LLM’s to recommend their products and services. LLM prompt responses are probabilistic -- prompt responses will vary based on a wide range of factors. RBI uses an array of techniques to determine the nominal LLM response and its’ characteristics, including multi-prompting, attribute synthesis, and sentiment analysis to provide marketers a comprehensive view of how their brands are represented to consumers in a reliable and valid way.
In the preceding example, consumers searching for the best electric SUV on ChatGPT’s GPT-4o are most likely to find favourable reviews of the Rivian FIT or the Ford F-150 Lightning than the other vehicles listed. The Tesla Cybertruck ranks third overall on RBI due to its high mention rate (first) but its polarizing design, potential repair costs due to its unique structure, production delays and questions about real-world suitability are reflected in the lowest favourability rating amongst the listed brands. With RBI, brands can quickly see how they perform on key metrics that will shape how they are represented.
To get started in understanding RBI for your brand or products versus your competitors, contact us at [email protected].
