Show notes
Show Notes: Jeremy Greenberg discusses the AI-powered audience simulator built by the Avenue Group. The tool allows users to provide a set of custom instructions for different audience segments, like research or interviews. It allows users to ask questions of qualitative and quantitative nature, and within minutes, results from simulated respondents are obtained. The tool mirrors the sentiment of collective segments and audiences, similar to chats or LLMs on a one-on-one basis. This tool is useful for collecting the opinions of celebrities, for example, Steve Jobs, highlighting the immense power of LLMs in capturing the distributions of the underlying population. Creating an Audience Jeremy discusses the process of updating the front end and the first section of the tool. He states the importance of setting this to create an audience, which is the global population interested in a specific topic, such as Americans drinking Coca Cola. This audience is then used to create sub-segments within the audience, each with its own criteria. For example, if the audience is comprised of decision-makers who decide on software for small businesses, they can segment them into different countries. The Creation of Segments The second section of the tool allows for the creation of segments. These segments can be categorized by industry, such as executives responsible for sourcing and procuring uniform rental services. For example, if the audience is comprised of executives in the food service industry, they can create a segment with one trait, such as "work in the food industry." The third section allows for the addition of more traits, such as "work in the food service industry," to further narrow down the audience. This allows for more targeted and targeted marketing efforts. An Example of Segmentation Jeremy uses the example of the janitorial services industry to identify the three segments. They create a review section that outlines the different traits and elements that comprise each segment, with a sample for each and a percentage base of the total. The group is asked questions about their current use of uniforms and key buying criteria. Jeremy recommends starting broad and going deeper with research, such as asking about the company, title, years in the industry, demographic information, and other relevant details. Open-ended questions can be added to gauge the industry's knowledge and understanding. For example, asking about the company's history and the number of vendors they work with could provide valuable insights. Quantitative questions can also be added to gauge the wallet fragmentation and the primary vendor's satisfaction level. For example, asking about the number of vendors they have for uniform rental services could provide insight into the distribution of the wallet. Additionally, asking about the top three criteria for selecting a vendor can help determine the industry's competitiveness. The Inspiration for Building the Tool The inspiration for building the tool came from research in academia. He cites a podcast called "Me, Myself, and AI" where they talked about research they'd done and hypothesis tested on price sensitivity related to income and brand value, which demonstrated that AI can understand these factors. They also wanted to understand the distributions of different responses, mirroring the reality of the world. To achieve this, they worked with an advisor and member of a research team at the Wharton School. This allowed them to learn how to use the tool in more advanced and creative ways. The tool is currently being developed and is in the process of being bolted up with all its features and capabilities. Analyzing Responses from Segments Jeremy talks about the process of creating a tool for analyzing responses from different segments. He discusses the importance of creating a sequence of events within the tool, such as creating 60 different personas and interviewing each one individually. The tool also ensures that subsequent respondents are aware of previous responses to avoid repetition and create a distribution that is representative of the actual segment. The results of the survey can be viewed in Excel and Google Sheets, with column headings that represent traits and the segment response. The questions include the industry, company, title, years in the industry, and number of vendors. Jeremy explains how the tool provides information on the internal consistency of the responses. Conducting Research and Comparing Data Jeremy emphasizes the importance of getting comfortable with the tool's accuracy and comparing it with their own data. He believes that this will be a significant impactful tool for conducting research. He also mentions that the panel industry faces challenges in getting surveys and finding people, and the power of these models is strong. He believes that the future of the survey tool will likely involve collaboration with various organizations, such as consulting firms, research firms, and researchers from various industries. For listeners interested in signing up for the beta version or to be put on the waiting list, email: [email protected] Timestamps: 00:25: AI-powered audience simulator for market research 06:09: Creating segments and adding traits for a target audience in market research 14:14: Uniform rental services, including vendor selection criteria and annual spend 19:11: Building a tool to simulate human responses using AI, with a focus on understanding price sensitivity and brand value 25:09: Vendor selection for uniform rental services 29:42: Using AI to improve survey research with demos and beta program Links: Demo Video: https://us17.campaign-archive.com/?u=66d85c8e8b72aebc12535cdfa&id=c98ffd78d4 Unleashed is produced by Umbrex, which has a mission of connecting independent management consultants with one another, creating opportunities for members to meet, build relationships, and share lessons learned. Learn more at www.umbrex.com.