C+R Research has a long history in segmentation research, going back to one of our founding partners, Steve Turner. Steve has been credited with being the first to pioneer occasion-based segmentation research back in 1992. To this day, we conduct many segmentation projects annually and are proud to be considered experts on the methodology.

Naturally, over the years, we have innovated our approach to conducting segmentation research while staying focused on sound methodology.  Many of the innovations have come from advances in statistical methods, which serve as the analytic “engine” behind good segmentation.  While solid methodology is the foundation, the proper socialization and communication of segment insights to stakeholders is the ultimate objective. Our combined quant and qual DNA has allowed us to provide powerful and engaging post-segmentation activation sessions to help bring the segments to life (such as speed dating and panel discussion with consumers who represent specific segments)– an essential step for client teams to get “up close and personal” with their target audiences to help develop empathy and leverage insights for identifying opportunities, driving innovation, and developing effective marketing tactics.

As soon as Large Language Models (LLMs) and LLM-powered platform providers became available last year, we immediately recognized the value and opportunity in using LLMs to create AI-based segment personas using data from our segmentation research. Our Artificial Intelligence (AI) task force set out to do just this.

Before jumping in, we created a mission brief for ourselves:

• The product needs to add value for clients, helping to bring segments to life and integrate segmentation insight into daily work conversations, quickly and iteratively. 
• The AI segment personas need to be deeply rooted in valid segmentation research.
• The tool or platform chosen needs to comply with our AI principles to guarantee reproducible, unbiased, secure, and privacy-compliant results.

After some initial testing, we decided to start with a segmentation that had particularly well-defined and differentiating segments. Our rationale is that the more distinct and separated the segments are, the easier it would be to assess the answers of the AI segment personas.  This would be our first “litmus test” – could AI representations of consumer segments accurately and reliably respond similarly to the way that real consumers would? 

A segmentation in the Vodka category not only yielded the well-defined segments we were looking for but also turned out to have relatable segments.  Well, relatable to some of us.

Shortly after we decided on the Vodka segmentation, we also narrowed the field of platform vendors to one platform (OpinioAI) to use.

Next, we iterated through approaches mapping the segmentation analysis plus supporting material into the respective AI personas. The results varied. We successfully created vodka-focused AI personas that had nuanced opinions about vodka, vodka consumption, occasions, mixology, etc., but these personas didn’t care about much else. Category-focused personas can provide a useful understanding of a category through a segment’s perspective, and we were able to validate category-focused AI-personas.

What are the ways that category-focused AI personas can be successfully used?  

We used the personas to evaluate fictitious product ideas, for example, a new brand of pre-mixed vodka cocktails. Our “Refined Vodka Lovers” segment persona expectedly recognized the ease of use of pre-mixed cocktails, but overall preferred mixing the cocktails themselves using high end spirits. On the other hand, our “Party Lovers” showed interest in the pre-mixed cocktails. So far, our experimental results show promise. We are also in the early stages of creating broader personas and personas less focused on the category. We are excited about the prospect of this and will provide updates as we progress. In the not-too-distant future, we envision creating “living” segmentation deliverables that include segment-tuned AI chatbots, which our clients will be able to use easily and cost effectively to leverage and infuse segment insights in routine development and refinement of strategies and tactics. 

Some of you may have already seen that C+R released a new, proprietary shopper-focused segmentation study With our newly developed AI-persona building knowledge, we’re eager to develop AI personas for our six shopper segments.  We’re starting to work on it this summer, so stay tuned for more news!

Check out our latest update on AI-based segment personas in our blog:

From Data to Personas: How We Are Harnessing AI for more Actionable Segmentation

---