A new look at personas: how UX combined with Data can help in the product funnel

How we brought our product team closer to our customers to start the relationship we want to have with them.

Two Nubank employees sitting in armchairs at the Nubank's office. They are working with their notebook on the lap, where you can see some stickers pasted on it.

Written by Nanda Lopes, with the contribution of Daniela Matos

Knowing the audience is an essential part of UX Design work. Understanding the different groups of people and their needs helps us build a product that better communicates with customers.

“Design establishes the relationship between your company and your consumers.”

Robert Brunner and Stewart Emery in the book Strategic Design Management

As the above phrase states, the relationship is a fundamental part of design and UX (user experience) work. It is by creating a relationship that we emotionally connect with our customers and generate an experience that not only satisfies, but creates fans.

To create a relationship, even in real life between 2 people, it is necessary to know the other person, their tastes and common things to create a bond. But if anyone can be a customer, as happens with Nubank, how do we begin to create this relationship? (Liking purple can’t be the only thing we have in common).

In this article I will share my experience of how we brought our Product Team closer to our customers. In addition to starting a relationship, the objective was to understand their needs in order to work on introducing a new solution concept.

Business problem as the starting point of the product

Until 2020, more than half of the people who registered on the Nubank app in search of a credit card, did not get the product. This happened because our credit analyses were very conservative and understood that approving credit for these people would have an impact on more risk for the business.

On the other hand, 44% of credit access requests in Brazil are denied for low-income people, according to a survey

It was from the union between the useful and the pleasant that the “Build Limit” function was born: on the one hand, we would like to increase the number of Nubank credit card customers; on the other, we had a sea of people who did not have access to credit in Brazil — and we knew we could help with this problem!

We knew who the target audience was, but we didn’t know the people yet.

We could say that the target audience of the Build Limit Function was all people who have little (or no) access to credit. But who are these people? What do they need a credit card for? Do they see value in the solution we are proposing?

Next step: creating personas convincing leadership about the need to know people better

For those who study design or marketing, building personas is part of this process of starting or improving communication about a product.

“They help us to focus on what is most important for our users and to put ourselves in the place of customers when making design decisions. Therefore, they should always be based on a qualitative study and reflect what motivates them”

3 Persona Types: Lightweight, Qualitative and Statistical from Nielsen Norman Group

The problem we faced was, in part, because there were already personas of Nubank customers. But these were not specific about credit card customers, much less about Build Limit Function customers or those who tend to be denied credit. 

Even though they helped Nubank to understand its audience in a macro way, they did not apply to the context of this new product.

We weren’t able to extract changes or practical information that would direct which problems we should solve, and what kind of improvements we should make.

“Those who have seen personas fall into oblivion without causing any significant impact on a general project usually discard them as a silly waste of time”

Why Personas Fail from Nielsen Norman Group

And those who think that this was a problem exclusive to our team are mistaken. Let’s do our role as designers and put ourselves in the place of leaders and Product Managers:

“To know the customer, for God’s sake!” – This is usually the standard response of designers (given in a bored and impatient way), because we have as the basis of our profession the need to put the customer at the center of all work. But to deal with the situation, to speak the language of the PMs and leaders’ audience and to convince them, we usually need data, numbers and metrics. And that’s exactly how we started.

Segmenting customers based on data (and a little creativity)

At this point in the product, it already existed for at least a year and we had already conducted research that helped us understand some pain points better. Our audience was no longer completely unknown and the product acquisition/conversion funnel was stable enough for us to understand what was normal.

We tried to segment our customers using demographic data and product usage according to our funnel. This step required a strong partnership with the Business Analysts on the team, who are the real experts in data analysis. Interestingly, none of the breakdowns we made to try to create groups of customers had statistical relevance, that is, we did not find different groups that would help us understand why one customer profile used the product more than another.

At this point, we need to bring in creativity and back it up with the results of previous research. What have customers already told us that could be an indication of the reason for wanting a credit card?

We arrived at the following:

Looking at these customer objectives, even if still as a hypothesis on our part, brought us closer to the person who was using or not using the product. And this hypothesis prompted further detailed data analysis that, ultimately, gave us a segmentation of customers that consistently had different behaviors from each other with regard to the product.

With separate funnels for each customer group, the entire team began to generate various hypotheses, especially leadership.

That’s when we received the green light to do a qualitative research that would bring us these answers.

Our research process

Now that the business was serious, nothing better than having another specialist leading the research. Our UX Researcher Daniela Matos took on the challenge and put together a research plan that ensured that the process was a good investment for the company and the team. In this section she will describe how it all happened:

To conduct the research, we needed to keep two things in mind:

  • Use a mixed methods approach (quantitative and qualitative) to complement our learnings;
  • Study the profiles separately, to not give a chance of mixing insights.

So far, we had discovered 4 groups of users, but delving into all of them would take a lot of time. In this way, the data team analyzed which of them were more strategic for our objectives. We selected 2 groups of users with the following criteria:

  • Number of representatives in our base;
  • Best results in the conversion funnel

That is, quantity + quality

The next step was to create a roadmap that could deliver insights quickly and test agilely. The format was as follows:

While the design team worked on the qualitative research with the 2 groups of users, the Business Analysts team provided us with support by generating databases and making complementary analyzes that would help us to support insights obtained in the interviews. Meanwhile the product team was testing changes in the product itself according to the insights obtained.

Data triangulation

As seen in the roadmap, we decided to research and test at the same time. This ensured that the research brought actionable items and a final solution that would actually have an impact on the product. In order for this to happen, we needed to define our hypotheses for each phase of the research.

In the end, the research results combined:

  • The learnings from Phase 1;
  • 23 hours of qualitative interviews from Phase 2;
  • Conversion and engagement metrics in the funnel analyzed in Phase 3;
  • NPS by user group analyzed in Phase 3;
  • Complementary data such as demographics, income distribution, credit card usage behavior, debt history, and account activities also seen in Phase 3.

Research Deliverables

At the end of this process, we were able to understand what people think, what motivates them financially and what behaviors they have.

We generated several reports, each focused on an internal audience such as leadership, product team, design team, among others, and today we have a roadmap based on these results.

Start by indicating the product best suited to the customer’s needs

Instead of creating personas with a photo taken from the internet or a fictional name, we were able to put side by side critical differences that completely change the way customers approach the Build Limit Function.

We understand that the product serves our audience well, but the value that should be communicated to these groups of people is completely different, as well as the goals of these people to have a credit card. 

Only then is it possible to create a relationship through UX, which makes that customer identify with what is being said and with the solution proposed by the product, which seems to really solve their problem.


  1. Get to know the people and their goals, not just your target audience. This improves depth and brings companies and people closer together.
  2. Work together. Have allies who are better at analyzing data than you and use them in all stages;
  3. Go beyond demographic data. For example, two people of different age ranges may have the same needs and behaviors;
  4. Different people absorb information in different ways. Find the way that works best for your team;
  5. Testing while doing user research speeds up learning. From the first studies (where we understood the difference between value propositions for profiles), we started to do experiments and think about benefits and offers that could make sense. The results already showed us the best ways, this way we maximize the time invested.
  6. Data triangulation helped us understand that we had more than one point of view on the results. A qualitative insight that had quantitative evidence had more weight. So, despite having many learnings, we were able to communicate concisely what really mattered.

        Design is also a strategy. Our challenge is to build the path between understanding the pain and needs of users and what’s best for the business. After all, Design is also strategy 😉

        Enter your name