The rapid rise in artificial intelligence (AI) in recent years, especially in the past few months, has deeply transformed our behaviours and routines. Given this technological revolution, adaptation has become an essential skill, and user experience (UX) and customer experience (CX) specialists are no exceptions to the rule.

You might be asking yourself things like “How is AI used in UX or CX?” or “Will AI transform UX or CX?” While the answers to these questions will quickly evolve in the coming years, I might have a partial response for you now—it goes by the name of “synthetic users.”

If you perform user research or use qualitative data on a daily basis, keep reading to get an overview of the AI tool Synthetic Users.

Synthetic Users: An artificial intelligence tool for user research

A few months ago, the CX, Data and Intelligence team at adviso tested the AI tool Synthetic Users as part of an actual client initiative. Created in 2019, the Portuguese company has taken off in the past year by specializing in user research… without any users!

Yes, you read that right: user research without users. Let me explain.

Synthetic Users is an AI tool that generates interviews and surveys, like in traditional user research, except in this case the process is completely simulated. The tool creates fictional users with detailed personality profiles built on data. Those users then interact within a simulated environment, make decisions, and evolve over time.

Credit: Sythetic Users 

At the moment, five types of simulated interviews are available.

1. Custom script: Ask your own questions—this is ideal if you have already prepared an interview guide.
2. Problem exploration: Analyze the points of friction and needs of an audience in order to identify opportunities.
3. Concept testing: Validate a concept with a specific audience. 
4. Ethnographic research interview: Study people from the subject’s point of view in order to understand their everyday lives and behaviours.
5. Research goal: Define your research goal and let the multi-agent architecture guide your interviews—this is the method that we’ve explored so far.

How user interviews with artificial intelligence (AI) work

Goal

To begin your research by goal, you will need to determine what you would like to look for. The research goal combines the subject, questions, and topics to be addressed. You can be as specific as you like, or not. It’s up to you!

Audience

You will also need a detailed audience with information such as:

• age;
• sex;
• occupation;
• location;
• spoken language;
• interests;
• salary; and
• other relevant information.

My recommendation is to work together with your colleagues in Insights to generate a well-structured, relevant audience.

Once you have completed this information, it’s time for the platform to get to work. Synthetic Users will generate user profiles in response to your requested audience. Each profile will have its own characteristics and personality. You can refine your research by adding supplementary parameters. For example, certain users could own a certain product while others don’t so that you can see how their opinions diverge.

Additional reading: The Ultimate Guide to Customer Loyalty Through User Research

Interviews

The interviews are launched. This is the time to analyze these “conversations” and decide whether or not to add your own questions. Our artificial intelligence experts recommend asking a few complementary questions to enrich the study while ensuring that you are not only collecting information from a single source.

If you’re thinking, “Wow, all I have to do is read the interviews and create a summary!” get ready for a shock—Synthetic Users handles that part too! The AI tool generates a document that summarizes and consolidates all the concepts presented while even adding key citations and potential avenues for further reflection towards the creation of future questions.

An audience report created by our CX, Data and Intelligence team using information generated by Synthetic Users

Fast results and illuminating data

In just a few hours—reading the interviews takes the longest amount of time—Synthetic Users allows you to learn more about the motivations, needs, and perceptions of users. According to the company’s own tests, 85 percent of the information obtained during user research with humans overlaps with that provided by synthetic users.

The data collected let you make informed decisions and optimize the strategies you have in place. They notably offer a better understanding of a clientele and their positioning within an industry, and at lighting speed. For companies or sectors with insights that are limited or lack depth, this information can also supplement a media plan and provide a super-precise demographic profile.

In addition, these qualitative data can be exploited for A/B tests and personalization strategies. Whether you’re generating test ideas, identifying points of friction in the customer journey, choosing visuals, or refining the tone of content, the testing phase has become a key lever for qualifying and validating a hypothesis.

Bias and the limitations of artificial intelligence in user research

It’s obvious that this type of tool has biases as well as potential limitations in order to frame this type of qualitative research. Here’s how to deal with the most significant of these limitations:

• For the purposes of representation and privacy, use a minimum audience population of 100,000 people. While this is an estimate, if your audience is significantly smaller, potential bias could be more of an issue. Since the fictitious users are created from available data, research that is too precise could enable the people who provided the information to be identified.
• These data should not be the sole focus of analyses. To make strategic decisions or determine an approach, complement your research with quantitative and behavioural data as well as other user interviews in order to obtain more nuanced information.
• Lastly, as mentioned earlier, round out your study with your own questions in order to obtain a reliable perspective on the information you’ve collected.

Synthetic Users vs. ChatGPT

Why not just ask ChatGPT? Technically, Synthetic Users makes use of the ChatGPT AI software. The information is simply in an off-the-shelf format. It should be possible to obtain these kinds of results in ChatGPT by employing several queries. However, it’s a lot easier and faster to use Synthetic Users, which is continuously improving its approach.

Will AI transform CX and UX?

The important takeaway from all of this information is that qualitative user research as pursued traditionally remains important and highly valuable. On the other hand, thanks to artificial intelligence and the creation of synthetic users, user research has become easier to integrate into everyday strategy development given its impressive time efficiency. That means it’s important to learn how to use AI tools such as Synthetic Users in a smart, nuanced way!

This type of platform is still in its infancy, but it shows promise for exponential development in the coming months. By adopting it today, you’ll be at the forefront of the fascinating, inspiring changes that are occurring.

At adviso, artificial intelligence supports our work every day! We can help you incorporate AI into your strategic or operational processes. Get in touch with us to talk about it.