Understanding customer needs is becoming more challenging—and more critical—by the day. As markets and customer needs evolve, organizations must stay ahead of the curve by identifying and addressing emerging customer needs before their competitors do. How can organizations successfully capture those unmet needs, and turn them into actionable insights ripe for innovation? More importantly, how can they turn customer feedback into intelligence that fuels innovation and differentiates them from the competition? The secret lies in leveraging AI-powered Voice of the Customer (VOC) research.
Why use AI for VOC research?
Today, many organizations are sitting on a goldmine of customer data—both proprietary and publicly available. Proprietary data might include customer interview transcripts, sales call notes, or call center data. Publicly available data typically includes user-generated content such as social media posts, online forum discussions, or online product reviews. Due to the explosion of available information, identifying customer needs from these sources has become time-consuming and overwhelming.
In collaboration with researchers from MIT Sloan School of Management and Northwestern University Kellogg School of Management, AMS sought to explore the capabilities of AI beyond its obvious ability to sift through mountains of data. The main research question: could AI identify and formulate detailed customer needs?
The answer? A resounding yes.
Applied Marketing Science (AMS) is thrilled to unveil our novel supervised, finetuned LLM, tailored for in-depth VOC analysis and capable of developing structured needs statements with human-like nuance, empathy, and contextual awareness.
How does the supervised, finetuned LLM work?
This unique model—the successor to AMS’s 2018 machine learning algorithm, ACE™—was trained using a comprehensive panel of VOC study data spanning numerous product and service categories including activewear and apparel, lawncare equipment, sleep aids, telehealth, hearing aids, and glucose monitoring. The supervised, finetuned approach mimics how professional researchers learn to analyze data: by studying real-world examples and iteratively refining their output with feedback provided by expert consultants. Where ACE™ was capable of pulling out insightful quotes from research datasets, the supervised, finetuned LLM is capable of fully formulating detailed customer needs on par with those generated by professional market researchers.
The model aggregates multiple data sources and research questions into a single study, captures nuances, unveils emerging trends, and formulates detailed needs statements. The team of researchers even found that the LLM is able to uncover needs that other methods of analysis were not able to unearth – in one instance, it was able to identify 60 tertiary needs that were initially overlooked.
Where traditional methods require interviews and manual analysis, the AI model delivers rapid, scalable insights using existing data, without compromising on depth nor quality. The LLM is capable of not only identifying customer insights but also formulating structured needs statements with human-like nuance, including empathy and contextual awareness. Researchers hypothesized that the LLM would perform better with functional needs than emotional, but it performed well with both. The new AI model represents a significant advancement over AMS’s earlier model, which was limited to extracting insightful customer quotations.
By eliminating privacy concerns, accelerating insight generation, and uncovering findings that might otherwise remain hidden, this powerful, forward-looking tool is revolutionizing VOC practices. Learn more about our model with this quick overview video:
Ready to transform how you uncover important customer needs? Connect with one of our AI experts to discuss how the new LLM can supercharge your VOC efforts.
RELATED RESOURCES
Supercharge Your VOC: How Supervised Finetuned AI Models are Changing the Game [Webinar on Demand]