One possible use case for AI-supported research is competitor monitoring. For a client in the food sector, we compared vending-machine providers and developed a recommendation based on predefined criteria.
We first conducted the research manually and then compared the results with those produced by ChatGPT and Perplexity. Both tools added information to the comparison tables, although some of it was incorrect. Overall, ChatGPT was more precise: it provided additional details on one vending-machine model and its capacity that we had missed during the manual research. Some prices had also changed in the days following our initial analysis, which both AI models identified. However, ChatGPT also returned some information that could not be verified.
ChatGPT was also more convincing when it came to the recommendation. Without being explicitly instructed to do so, it structured the analysis around several criteria and provided practical suggestions for the client’s decision. In the new “Thinking” mode, fewer errors occurred and more information was provided.
Our conclusion: From a research perspective, ChatGPT 5 can be useful for quality-checking competitor analyses, particularly in “Thinking” mode, which can be selected with a subscription. However, all factual information should still be verified.
Our recommendation for this type of research: AI should ideally be used for competitor monitoring only after conducting independent research first. It remains relatively unreliable for facts and figures, but can add value in more creative tasks such as brainstorming or developing recommendations – even if that may initially seem counterintuitive.
By Maria Kleiner
