← Video Library · ChatGPT & GPTs
How to Create Interactive Customer Personas with AI
Published August 27, 2024 · 442 views on YouTube
Key takeaways
- A custom GPT can be configured to simulate multiple buyer personas at once, based on the market segments described in a buyer's brief.
- The starting prompt for the persona-simulator GPT was originally developed with Claude, then uploaded and refined inside the GPT configuration flow.
- Uploading a buyer's brief as reference material lets the GPT generate role, constraints, guidelines, and prompt starters tailored to each persona.
- Once built, you can converse with the simulated personas directly, asking things like what they would value about an event or what they would dislike, and get in-character responses with reasoning.
- Feeding the GPT more real customer data and conversations over time, beyond just the initial buyer's brief, deepens and refines the personas it simulates.
How do you create interactive customer personas with AI?
Build a custom GPT and give it a buyer’s brief as reference material, using a prompt (developed with Claude) that instructs it to act as a simulation of the different buyer personas described in that brief. The GPT then generates role, constraints, guidelines, and prompt starters for each persona, and you can converse with those simulated personas directly inside the chat.
What do you need before setting up the GPT?
You need a buyer’s brief, a document describing your target market segments and the personas within them, and a starting prompt written to instruct a GPT to act as a simulation for those different personalities. In this example, the prompt was originally developed using Claude and a buyer’s brief example for the Side Hustle Game, then refined for use as a GPT configuration prompt.
How do you set up the persona-simulator GPT?
- Open the GPT dashboard inside your OpenAI account (accessible from your account settings) and start a new GPT conversation.
- Provide the prompt developed for simulating buyer personas as the initial context for the GPT.
- Upload the buyer’s brief document along with that first prompt so the GPT has the persona and market-segment details to work from.
- Let the GPT generate its role, constraints, guidelines, clarification, and personalization based on the context provided, then review and adjust as needed.
- Move to the Configure tab to confirm the description and prompt starters the GPT generated, and add any additional context, such as referencing the buyer’s brief directly for deeper understanding of each persona’s values and perspective.
- Click View GPT to start interacting with the simulated personas.
What can you actually do with the finished GPT?
Once built, the GPT responds in character as the different personas from the buyer’s brief, for example a “business strategist” persona, and you can direct the conversation toward any of them. Example prompts include asking what content would resonate with a given persona, or what that persona would dislike about an event and why, with the GPT citing reasoning tied to the persona’s perspective, such as concerns about a lack of clear structure or feeling creatively stifled.
How do you make the personas more accurate over time?
The buyer’s brief is just a starting point. Adding more real customer data, such as discovery call notes, past conversations, or details about the buyer’s journey (what has worked and what hasn’t), to the GPT’s knowledge base makes the personas it simulates more accurate and useful for that audience segment.
Full video transcript
In this video I’m going to share how you can relatively quickly get a basic GPT simulation set up for multiple buyer personas. First I’ll open up my GPT dashboard inside my OpenAI account, which you get to by navigating to your settings, and start a new conversation of the chat.
What I’m going to be making this based off is actually a prompt that I developed with Claude, using a buyer’s brief example for the Side Hustle Game, and I did a few quick optimizations to basically create a prompt that has it act as a simulation for different personalities based on the market segments.
I’m going to take this prompt to be the context, and I’m going to actually upload the buyer’s brief as well with that first prompt to get the process going. I’ll go back to Configure and modify the prompt based on the direction it goes, calling this a persona simulator because I’m going to create it specifically for this buyer’s brief.
Notice how, now that I’ve given it this basic context, it’s going to give me the role, constraints, guidelines, clarification, and personalization, and update the instructions to include this. Sometimes it’ll miss context or give back too much context, but in this case the guidelines look like they’re being optimized based on what I provided it. So for prompt starters, I’m just going to say okay, and it’s going to update the GPT.
While it’s updating this GPT, I’ll be able to move over to Configure and see how it’s actually getting things set up, and add additional context and information from there so you can start to simulate the interactions. I’ll go back over to Configure and you can see how it updated the dynamics, and when appropriate, the description. I think it actually gave the prompt starters, and it included a demo for Side Hustle. I’m going to add a reference: demo Side Hustle to buyer’s brief, for deep understanding of each persona and their perspective, as it relates to what they value and what they want to get from the brand.
I’ll pause it here. You can see it’s updating in the top right-hand corner. I’m just going to update it myself, and then click View GPT. So now that I’m here viewing my GPT, if I’m acting like this is embodied based in the Side Hustle Game, here’s the overview of the personas I embody. You can see here it’s taking on the role of these personas, and we can guide and engage inside of this chat GPT simulation however we want to steer the conversation. We could customize the knowledge base with specific context, additional works, and conversations from individual customers, to refine their perspectives.
As a test, I asked about considering doing an in-person event, just a basic question. It’s now responding with the different personas, like a business strategist, and from this point you can pick through which one you want to talk with. You could ask, tell me more, what are some types of content that would resonate with you doing X, Y, or Z. Another example question: what would you hate about the event, give reasons and cite sources. It goes back into their perspective: lack of clear structure and organization, feeling overwhelmed, a competitive atmosphere stifling creativity.
You could take this in almost any direction you want, and the more context you have, the more you can do. This was just the buyer’s brief as a starting point. The more customer data and information about a buyer’s journey, what’s worked and what hasn’t, the more that can be used for the knowledge base for that particular audience segment, and all the different personas inside a brand might be reaching would be accessible here in an interactive nature. Once you have the GPT created, you can go back and edit it and build on it from there.
I hope you find this use case helpful. If you have customer data, or just want to take what you’ve created with a blueprint based on research and information and add that to the knowledge base, it’s a way to actionize content from just discovery calls with your customers or clients, to brainstorming ideas and perspectives from AI versions of your ideal target audience. I hope you find this helpful, and I’ll see you in the next video.