Generative AI that produces accurate market research? It’s true. In the first half of 2023, Harvard Business School Marketing Unit published a paper on AI use cases for market research. The authors found that results generated by large language models (LLMs) like GPT were both realistic and aligned with economic theory and prior research. They concluded that “GPT, and LLMs more broadly, can serve as a powerful tool for understanding customer preferences.” Just a few months later, iSoftStone’s engineering services team was helping a global technology company leverage Azure OpenAI for market research.
Like other companies that make hardware and software products, our client develops careful go-to-market and product development strategies with insights from third-party research and statistical analyses. These insights are vital, but also time-consuming and expensive to produce. Our client was curious if generative AI could help them reduce or mitigate the overall cost of market research.
Revolutionizing conjoint analysis with Azure OpenAI
The first proof of concept (POC) we developed was a conjoint analysis tool in which the AI “played” the role of consumers and responded to a survey. Our engineers built the matrix and logic for the tool, refining it for efficiency and partnering with our client as they iterated on the prompts. Since natural language answers aren’t required for a conjoint analysis, we used Azure OpenAI completion endpoint deploying in GPT-3 and GPT-3.5. The goal was firstly to see if the tool could identify product attributes seen as highly valuable by consumers and, secondly, to explore if the results tracked to the previous year’s (traditionally gathered) conjoint analysis.
The tool’s results did indeed track. And, importantly, our client realized the potential volume of information they could generate was greater. Unlike a traditionally gathered conjoint analysis, there’s no upper limit to the number of product variables you can ask an AI to review or personas you can ask it to take on.
AI-driven consumer willingness analysis
The second AI for market research POC we created was a tool to analyze consumer willingness to purchase new products. It assesses variables like price points and service features to identify purchasing triggers and inform our client’s go-to-market strategy. We scoped the tool with a range of different B2B personas — for example hypothetical clients with different industry sizes, number of employees, and years of experience — and consumer scenarios. The AI would report back on willingness to buy. Again, the results were in line with our client’s traditionally gathered third-party research.
It looks increasingly likely that AI tools will enable market researchers to do more in-house and deliver ancillary benefits like cost saving. For now, our client continues to run traditional third-party research while they work with us to refine the POC tools and experiment with the limits of Azure OpenAI. It’s still early days, but they’re excited by the possibilities ahead.
What’s your AI vision?
From market research to medicine, content generation to chatbots, all sorts of industries and organizations are using AI to transform their operations and customer experiences. Ideas that used to be purely theoretical or futurist thinking are rapidly becoming business reality for our enterprise clients.
As a Microsoft Gold Partner, iSoftStone is delighted to see Azure OpenAI Service helping more businesses innovate fast, solve big problems, and deliver savings. In addition to our extensive knowledge of Azure OpenAI, we bring 15+ years of strategic partnership with Microsoft as a supplier and cloud solutions provider. We’ll help you brainstorm, define your AI vision, and build a compelling POC that brings it to life.
