How I used ChatGPT for user Personas
Here’s my experience
User Personas those fictional representations of your ideal customers that help you understand things like their needs, goals, and pain points. But creating them can be a real pain, especially if you don’t have much data or insight about your target audience. Funnily enough, I knew who my target audience was but had NO idea on the kind of characteristics they embody.
How I approached personas
As one would, I started out with demographic data such as Name, Age, Gender, Occupation, etc. This was the easy part. I could come up with a different fictional character each time with corresponding occupation.
Next, I had to come up with personality traits: likes and dislikes, online presence, and motivation. This is where things got chaotic. With my limited knowledge of how to fit the puzzle pieces, being everything else other than demographic data, it broke my head to try and come up with something convincing.
I based everything on my own knowledge and whatever I could Google. Being unbiased was important, because I didn’t want my fictional users to be too fictional. Here’s an example:
GPT Personas
I was initially skeptical about using Chat GPT for creating personas, maybe because I’d never fully used it to it’s potential.
I stared out by asking what goes into a persona to understand the kind of input that’s required. This helped form my next prompt.
Since I had already identified the target audience, I could use that information and feed it into Chat GPT for context.
But I wasn’t entirely satisfied with the result.
Then, with the help of Rohan Sathish we decided to make our own model for Chat GPT so it gives us a very specific output. Here’s what one of the personas look like (Couldn’t show the chat as the information was getting cut off):
You can see, that with a bit of prompt engineering you can get pretty amazing results from Chat GPT, and the best part is that it also remembers context.
Since the GPT model has been trained on an unfathomable mountain of data, it’s giving a more accurate representation about the characteristics of our target audience. This drastically reduces the amount of guess-work that goes into making these personas.
How Much Time I would’ve saved
My task was to make 15 user personas (yes I know it’s a lot, but it was necessary, trust me). I had to come up with 15 fictional characters most suited our target demographic and it took me around 3 days to complete.
With the help of Chat GPT, it took me 30 minutes to come up with 20 different user personas. 30 minutes.
I would’ve saved three whole days.
Should you solely rely on Chat GPT for creating personas?
As with any AI tool, it’s meant to only help you come up with more ways of doing the same thing. Chat GPT can only help you if you know what kind of data you require from it. For me, since I had all the information ready about the users it was easier to provide context to it. On the contrary if you ever need to generate a set of target audience for any purpose GPT is still an amazing tool, provided you give it enough context.
In Conclusion
Chat GPT for personas? Definite yes. Chat GPT for anything else? Absolutely. Hotel? Trivago.
Chat GPT is an amazing tool that can aid you in your research and discovery phase of design, it’s not meant to replace the entire process in itself. Having the right kind of knowledge about our target audience made it a whole lot easier to prompt Chat GPT to get the specific output that we needed.
The kind of response it’s capable of giving opens up a whole new way of working and learning and it shouldn’t be ignored. We have to use it to be better, and to unlock new dimensions of thought.
“AI is taking our jobs”. Not quite.
If you want to check out how we actually pulled this off read this article: Using ChatGPT To Accelerate Design
