Brand planning, ChatGPT & Me

The following is a summary of an online conversation. It took place over several hours, over the course of a few days, with ChatGPT4. This single, continuous conversation started with me sharing transcripts of customer interviews we had conducted, then progressed through 3 qualitative analysis phases, to finally arrive at some draft positioning options for a client’s brand.

The purpose of reporting this conversation is not as a teaching. I’m a LLM novice and I’m sure there are better ways to prompt engineer this kind of interaction. Rather, my goal here is to document my personal experience of an experimental collaboration between Thinking Human and Generative AI in the development of a healthcare brand’s positioning. To protect client confidentiality, I’ve included my initial message in each phase of the conversation, but just described ChatGPT4’s responses. I would like to thank David Boyle and Intellus for getting us started with their great Webinar earlier this year.

Phase 1: Market landscape

Me: I am a market research analyst working on healthcare brand X. Please read all the interview transcripts with [target customers] from [5 countries], in the consolidated document uploaded. Based on these interviews please summarize

·      Attitudes, perceptions, challenges, trends related to the [category definition]

·      Functional and emotional needs when evaluating brands in the category

·      Associations and perceived strengths and weaknesses of the brands

·      Key drivers of brand choice

Bring the summary to life with direct quotes from [target customers]. Highlight any key differences between the countries.

ChatGPT4 couldn’t handle the entire brief in one go, so after some back and forth, we completed the 4 sections one by one. It appeared to have nailed the summaries at the first attempt, but then I realized it had totally ignored some of countries. Only after I insisted several times that it read all the interviews, did ChatGPT4 acknowledge country differences in its conclusions. I asked follow-up questions to get at the ‘whys’ more. It elaborated as best it could, informing me when it was “conceptualizing beyond the transcripts”. I shared my final insights summary and key drivers of brand choice in the next phase briefing message.

Phase 2: Needs-based Segmentation

Me: Create distinct [target customer] segments based on their functional and emotional needs. For each segment, please define what the [target customers] are looking for from brands in the category and illustrate their choices with useful direct quotes and example brand choices.

We tried this segmentation exercise several times, adjusting the instructions to see what emerged. ChatGPT4 generated a range of segmentations, from a version with 7 very creatively titled segments, to one with 3 more prosaic, descriptive ones. All were grounded in real functional and emotional needs derived from the customer interviews. In the end, I settled on a segmentation that made most intuitive sense to me, based on all the insights discussed so far. I included this proposed segmentation with the next briefing message.

Phase 3: Positioning

Me: For each segment, please propose a brand positioning for healthcare Brand X based on the following: [Brand X features and benefits list]. For each positioning statement, please include:

·      The needs-based customer segment we are targeting

·      The market frame of reference brand X competes in

·      The main differentiating benefit (rational and emotional)

·      The reasons to believe the main differentiating benefit

Explain how each brand positioning is truly differentiated from the following competitors: [Brand X competitors list].

ChatGPT4 created a very clear positioning statement for each target segment. It went off to research more background on the competitors when there were no mentions in the interview transcripts, so it could better “conceptualize the differentiation”. It gave links to the source webpages when doing so. I re-briefed ChatGPT4 several times, providing more background on the competitor brands and tweaking the Brand X features and benefits list, to create alternatives. Some versions seemed a little naïve, in others ChatGPT4 showed remarkable insight into Brand X, its target customers and differentiating benefits. Finally, I crafted the language some more (robot stigma makes me feel I have to say that), and put it into PPT.

To Sum Up

My take on this collaboration was that it was like working with a somewhat difficult, genius student, who knows everything, but needs direction to achieve their potential. As a brand planner, ChatGPT4 is logical, strategically brilliant, surprisingly creative, and quickly learns what you want if you are clear and give sufficient context. Despite having to check ChatGPT4’s work, there were significant speed and efficiency benefits in the whole process. As with the genius student, to me the biggest benefit of working with ChatGPT4 is the inspiration and ideas offered up for consideration. The interaction shaped and sharpened both our thinking.

It was good, almost as good as the real thing.