All eyes are now on ChatGPT, the AI tool that managed to hit the one million users mark just five days after its launch. Being totally free, but very effective and able to provide tailored answers to users’ questions, ChatGPT is becoming an integral part of our daily lives, both personal and professional.
Here at Eleken, we’ve already tested ChatGPT and agreed that it is quite useful for doing UX research. Based on our experience, the tool offers rather accurate and realistic responses to a wide range of inputs. But we also have to say that the tool comes with certain flaws, which we’re going to discuss later on.
In this article, we’ll take a look at the benefits and limitations of using ChatGPT. We will also provide some examples and tips when you want to use it for your project, so keep reading.
Benefits of using ChatGPT for UX research
Conducting UX research not only helps designers develop products that appeal to future users but also ensures the designed solutions will be competitive. But the truth is, UX research is a time-consuming and complex process.
According to the State of User Research survey, the typical study takes from 1 to 4 weeks to plan, while 61% of respondents admit that they conduct half of their research sessions with participants from their own audience. Such research practice may lead to inaccurate results and bring bias in an assessment of market needs.
In this respect, ChatGPT can be very helpful for speeding up the process and providing unbiased opinion. As a learning-based natural language processing tool, it is trained on a diverse range of text data sourced from the Internet in a variety of ways.
The use cases for ChatGPT in UX research are endless, starting from gathering inspiration and ideas to providing a color palette for an app or a website. The tool can help develop the concept and recommend fonts or an icon for a non-standard project.
With that said, the benefits of using ChatGPT for UX research include:
- Natural, human-like performance. Based on the recent GPT-4 Technical Report, ChatGPT can now understand both text and image inputs, and generate a wide range of responses in a close-to-human manner. The research showed promising results, demonstrating that ChatGPT can interact with human language and follow instructions creatively, like humans do, which is beneficial for UX researchers looking to collect user feedback.
- Great level of accuracy. The accuracy of GPT-4 has improved compared to previous models, achieving scores of over 80%. This means it performs better in predicting and generating more contextually appropriate responses. Simply put, the tool can now give better and more sensible responses based on the question asked.
- Multi-purpose. From text generation to question answering and building user personas, there are many ways how UX researchers can utilize ChatGPT.
- Cost-effectiveness. Lastly, ChatGPT is not only fast and accurate, but comes at no cost, letting you optimize the budget for UX research.
Okay, now it seems like we’re complete ChatGPT advocates and are about to encourage you to use the tool ASAP. But as we mentioned above, ChatGPT is not almighty, and we’re about to explain why.
The limitations of ChatGPT
Despite all the benefits, caution is also needed. ChatGPT sometimes generates gibberish responses. The tool still lacks knowledge of current events and can miss subtle details. Plus, the source of the information is unknown, which is why you can't trust it fully.
Apart from that, Stanford University names such areas of risk as cybersecurity and trustworthiness. There still exists the difficulty of eliciting bad behavior or the risk of generating harmful advice.
The OpenAI team admits that ChatGPT can’t fully replace humans, but rather serves as a helpful assistant. So, when you want to use ChatGPT for UX research, you will still have to validate the information generated by the tool and engage a UX professional who knows all the nuances of UX research.
Some examples of using ChatGPT
Here at Eleken, we find ChatGPT quite useful for conducting UX research. Here are some examples of when the tool can help, in particular:
Gathering market insights
The UX research process involves identifying competitors and understanding the market, gaining insights into user preferences and behavioral patterns. Gathered data helps understand how users may interact with the product we design.
Let’s say you need to collect information about the target market and potential users to verify the success of a new student engagement app. Here’s what ChatGPT recommends.
Prompt: I want to gain insights into user preferences and behavioral patterns to understand how they will interact with the student engagement app.
Additional prompt examples:
- Create a list of top competitors in the education market. Write their strengths and weaknesses.
- I want to know how to conduct primary market research to gain insights into customer needs and preferences for the education sector. How can I do it?
- How can I use secondary market research like the education sector competitor analysis to gain a better understanding of the market?
- What are effective ways to segment and target the education sector audience for market research?
- How can I effectively collect survey responses and conduct questionnaires to gather market research data?
- What are some best practices for conducting focus groups and user interviews to gain insights into customer behavior and preferences?
- Imagine you are a UX researcher. Write me a detailed UX research plan. Include JTBD statements, personas, and desired outcomes.
Helping create interview questions for qualitative user research
Doing both quantitative (qual) and qualitative (quant) UX research is essential for all kinds of projects. While qual research provides observational findings and allows you to understand users’ emotions and behaviors, quant research offers metrics and actual data. Sadly, ChatGPT can’t help with the latter much, but can it help us with qualitative research? Let’s find out.
Prompt: I want to improve the design of the student engagement app. How can I craft effective interview questions for qualitative research?
Additional prompt examples:
- What are the best ways to analyze and interpret qualitative research data?
- How can I use data visualization to present user research findings?
- How can I do a survey for quantitative research?
- What are some best practices for conducting surveys for student engagement apps?
Building buyer personas
UX researchers create personas to identify ideal users and assess design decisions. Let’s ask ChatGPT to build a user persona for the student engagement app.
Prompt: Build a user persona for a new student engagement app that automates the onboarding and arrival process for university students and agents.
ChatGPT created a user persona by the name of Sarah. For this persona, it specified age, background, goals, challenges, and needs. Additionally, the tool offered a list of key features for the app.
Additional prompt examples:
- What are the best practices for building user personas based on real user research and information?
- Create a persona for a time-tracking web application interested in managing their working time and getting insightful reports.
- How can I use user personas to improve the usability and user-centeredness of web applications?
A Non-Boring Guide to How UX Research Is Supposed to Work
Learn how to approach UX research to grow your product
Get the free eBook
Preparing interview questions
Once we’ve created user personas, the next step is to find users who match those criteria and interview them. Let’s ask ChatGPT to generate the right questions.
Prompt: I want to interview time-tracking web app users. I want to know what functionality they need more. Write 10 interview questions I can ask them.
Additional prompt examples:
- What are the best practices I can use when conducting user interviews?
- Generate questions for a research interview to gather feedback on a new student engagement app.
Selecting color palettes and features
ChatGPT can also be used to prioritize the features users expect to see in a product. But you can easily use it to pick up certain color palettes that appeal to the target audience or select complementary colors.
Let’s ask ChatGPT to suggest primary and secondary colors for a time-tracking web app and a list of features users want in the app.
Prompt: Can you suggest primary and secondary colors for a time-tracking web app? List features users expect to see in this app. Represent in a chart.
Additional prompt examples:
- Suggest 2 color palettes for the student engagement app.
- Can you create a list of features users want from a student engagement app?
- Can you prioritize key features and requirements for a time-tracking web app?
Tips to incorporate ChatGPT into UX research
Working with ChatGPT reduces time spent on UX research, making the process of gathering information easier. The researchers can leverage the capabilities of the tool with minimal effort. Here are some tips that can help:
- Ask simple, open-ended questions. To get more accurate results, it is better to use open-ended questions starting with "What," "How can I," "Why," "Can you explain", and such, instead of yes/no questions or the ones that provide a single word or phrase responses. Complex sentences, unusual words, and structures or technical jargon are best to avoid since they can confuse the chatbot.
- Mind the length. According to Arxiv, ChatGPT has about 4,096 of context length (that’s 3,072 or 6,144 words), so keep that in mind when you’re about to have open debates with the chatbot.
- Test it on your own. The best way to understand if ChatGPT fits your needs or not is to give it a try. Ask for custom responses, add context, and specify information about what you need – this way the tool can better adapt to your needs.
- Teach the bot. ChatGPT is an AI learning model, which means you can teach it and improve it. So when, for example, you don’t like the answer provided by the chatbot, you can regenerate the response and rate whether the result was better or worse.
Conclusion
ChatGPT offers great opportunities for UX researchers. But if you want to unlock the full potential of the technology, you have to spend a day or two learning how to use it.
Remember, no matter how good ChatGPT is, it’s still a machine and you can’t fully replace humans with it. So if you want to use ChatGPT for things like research, it would be wise to have a professional you can reach out to. And you can find them at Eleken, by the way.
If you want to learn more about UX research methods or need a consultation on what tools to use for UX research, the Eleken team is here to help you. Drop us a line here!
.png)
