How AI is enabling designers to go beyond personas

Personas has brought us this far, but it will no longer be enough for the approaching future.

Published in

UX Collective

8 min readAug 29, 2024

A meeting table seen from above with 8 people seated discussing ideas. — Give the Users a seat on the ideation table.

As human beings, we connect with stories. It’s much easier to remember a percentage when it’s associated with a use case or a person. In an era where every interaction is data, and even algorithms produce synthetic data, we often generate more information than we can analyze and understand.

Are we missing opportunities because we can’t identify value hidden in a sea of data?

This is the second article in a series investigating Artificial Intelligence and design research. In the first one, we discussed synthetic users. If you haven’t checked it out yet, it’s available at this link.

Here, once again, we go beyond the obvious without the pretense of offering definitive answers. The intention is to share valuable insights to stimulate reflection.

By the end of this read, you will have gone through the following topics:

Artificial Intelligence and the shifting of opportunities
Big Data and the ability to tell stories
From static segmentation to dynamic segmentation

Shall we?

Artificial Intelligence and the shifting of opportunities

First of all, it’s essential to understand that Artificial Intelligence causes a shift in opportunities.

The history of humanity is marked by countless technological revolutions that have radically transformed our lives. From the Industrial Revolution to today, new technologies always disrupt how things are done, often replacing ingrained habits and behaviors.

Take photography as an example: when cameras first appeared, many believed the profession of painters was under threat. Indeed, photography redefined the visual arts market but also created new professions — like photographers.

Whereas painters were previously hired to create images for posters and magazines, photography took over many of those functions.

However, we still see galleries filled with oil paintings. Just as today, anyone with a smartphone can take a picture, professional photographers are highly valued for their unique skills.

This same dynamic is present in the era of Artificial Intelligence.

Another example: Have you noticed the difference in quality between the images an artist and a person without that artistic background generates in Mid Journey? All the experience and background matter when writing a good prompt for Artificial Intelligence.

Collage of images: child lying on the floor, woman sitting at a bus door, woman in a dress with ants, and man in a suit jumping off a train. Created by artist Zack MacTavish using Mid Journey and inspired by renowned photographers. — Collage of images created by artist Zack MacTavish, inspired by renowned photographers, using Mid Journey. Source: Zack MacTavish.

The quality of the results depends on the input prompt, the database the algorithm has access to, how it was built, and the curation of the results to refine the process as a whole. The more we know what we want and the more intentional we are, the better the results we obtain.

This is true whether we’re generating an image with artificial intelligence, testing possibilities to create a new vaccine from algorithms, or coming up with a new business idea using ChatGPT.

But beware: this is not a frivolous reflection that overlooks the social and economic implications of Artificial Intelligence. It’s important to approach this discussion responsibly. Beyond a catastrophic view, I share here a perspective guided by the futurology perspective that urges us to make the future so desirable that it becomes inevitable.

Thus, it’s worth considering that technology generates a shift in opportunities: some are fulfilled by it, but others arise. Perhaps the secret is knowing how to identify them.

In the previous text, we discussed how Synthetic Users are neither purely quantitative nor qualitative research, but a third way. Today, we’ll talk about how they can help us surpass the results and methods we already know.

Just as painting and photography coexist, let’s reflect on this new universe that seems to be in front of us.

Just as it was difficult at the beginning of the internet to imagine everything that would happen today, the idea here is not to make any predictions about the outcomes of Artificial Intelligence. The idea is to ask questions and reflect more than provide answers.

After all, questions move the world and help us create more interesting realities.

One woman with a yellow shirt holding a pen and telling a story and two other woman in the background listening to it. — *In an increasingly dynamic universe, personas quickly become outdated.*

Big Data and the ability to tell stories

Have you ever heard that data is the new oil? Yet, according to Gartner, 85% of data science projects fail or don’t deliver expected results. I believe this could be related to our ability to tell stories.

We live in an era of data abundance. Every interaction we have, whether buying something online, watching a movie, or even ordering food delivery, generates data. And this data becomes the raw material for business ideas.

However, we often produce more data than we can analyze and understand. As human beings, we connect better with stories. It’s easier to remember a statistic when it’s associated with a person and their story.

That’s why, in design and marketing, it’s common to create personas, which represent the average profile of consumers based on their habits, desires, and needs. Usually, these personas are presented in slides or printed to serve as guides during business decision-making.

This strategy has been very important for businesses and is what brought us here, but everything indicates it will no longer be sufficient to handle new market disruptions. Consumer segmentations are more dynamic than ever, and behaviors change rapidly.

Personas quickly become outdated, and a slide or printed paper no longer suffices for what we need.

This is where new technologies, like Artificial Intelligence, come into play, paving the way for new approaches. But if we don’t use them creatively, we’ll end up with a universe of numbers that don’t evoke emotions or inspire action.

One alternative is Synthetic Users, algorithms capable of analyzing CRM data, social media, purchase history, and other company data, creating behavior groups in real time.

Synthetic Users, in essence, are a tool to help us create a narrative from a vast dataset.

This helps identify business opportunities that might go unnoticed amid an abundance of data. Of course, it’s important to talk to people and understand the reasons behind their behaviors. These new technologies don’t replace this approach but complement it, bringing new opportunities and expanding our understanding of the reality we live in.

After all, it’s no use having an infinite set of information if we don’t know the story it tells. That’s why the ability to build narratives that evoke emotions and engage is one of the main things that will help us on this journey.

Digital illustration showing different data graphics. — *When change is constant, being static means falling behind.*

From static segmentations to dynamic segmentations

In his talk about "The Death of Static Segmentations", Innovation Board Director Geoff Gibbins reflects on the new moment in user research. According to him, when change is constant, being static means falling behind.

New technologies powered by Artificial Intelligence allow for the real-time creation of dynamic, continuously improving user segmentations.

But how would this work in practice?

Imagine you are conducting a study for a large retail company that wants to understand its customers’ behavior to optimize the launch of new products. One common is to use a triangulation of methods, right?

One possibility is to study the information the company already has about its customers, whether static, like reports and previous studies, or dynamic, like CRM and social media interactions. With this information, you can look beyond your customer’s universe and understand what competitors are doing and how people are behaving in response to that, using benchmarking, for example.

From there, you’ll likely notice that there’s still missing information, and those pieces of the puzzle can only be obtained by talking to real customers. This is where the third stage of your triangulation begins, where you’ll recruit and talk to people, right?

After this process, you’ll have a wealth of data, and your challenge will be to tell a story from it that makes sense and helps the company address the challenge of optimizing the launch of new products. So you create a fantastic report, with personas and interesting ways to visualize the data. You identify average behaviors and present them to the company in a way that facilitates decision-making.

Some time later, the results would already be outdated, and it would be necessary to repeat the entire process to address a new business challenge that has arisen.

Sound familiar? Many projects have been conducted this way.

A meeting room with two pepole at front and seven people sitted, discussing ideas in a work environment. — *In a different scenario, the design deliverables change.*

What changes now is that after the presentation meeting, the study doesn’t necessarily have to end. Tools like Large Language Models allow us to simulate behaviors based on the data we’ve collected, making it possible, for example, for decision-makers to ask questions to a synthetic user in the same way they would ask you when you were presenting the project.

Beyond considering it as a research tool, it could actually be a way to make the research deliverables more interesting and appealing.

We know that has been a lot of debate around if, when, and how to use synthetic users in UX Research, maybe this could be an alternative.

We can plug in our dynamic data sources and keep enriching the data base with static reports and various data sources, creating an ongoing research knowledge management.

Before, the personas printed in a PDF stuck to the meeting room wall didn’t talk. The researcher answered the questions that arose in the meeting based on what they had learned with the users. Now, synthetic users can have conversations.

The most interesting part is that the UX Researcher’s and Designer’s role also changes: we are no longer just the client’s spokespersons within the company; we can also be responsible for curating input data, validating results, and creating and continuously improving a process that ensures data continues to be collected in a relevant way. And, of course, we remain the spokespersons for spreading these ideas within the company, creating a favorable environment for innovation.

And yes, like any other research method, we need to be aware of biases and how to address it. We can design the models to better handle this issue and keep improving it. As good design thinkers, we can have prototype a MVP and start from there, right?

For each business need, there is an opportunity for audience segmentation.

Once we establish a healthy, curated, real, and consistent data repository, we can dive into this sea of information according to each business need and identify specific audiences. After all, a brand can have various dynamic audiences.

Do you have questions or want to exchange more ideas about innovation? Feel free to send a message!