Maria Gomes - Product Designer

Fintech

B2C

Experimentation

Increasing sign-up conversion at Bitso, one experiment at a time

Bitso, 2026

MY ROLE

Product Designer

TIMELINE

January 2026

TEAM COMPOSITION

Product Designer, Product Manager, Content Designer and a Developer

ABOUT

Bitso is a financial app backed by crypto, where people in Latin America can buy, sell and hold digital assets.

When I joined, the Acquisition team didn't exist yet, and together with the PM, I helped define what we would build.

SUMMARY

Sign-up is where a user starts their journey with Bitso, yet nobody was paying attention to it. I took this challenge myself, audited the whole flow and turned the findings into the Acquisition team’s A/B testing roadmap.

Two improvements have already shipped, each lifting conversion by more than 2%. And after the first one, sign-ups grew by about 2,000 a month.

The biggest win wasn’t any single experiment, but a backlog with simple yet powerful experiments we could keep running.

Two improvements have already shipped, each lifting conversion by more than 2%. And after the first one, sign-ups grew by about 2,000 a month.

The biggest win wasn’t any single experiment, but a backlog with simple yet powerful experiments we could keep running.

Nudging users to complete the email verification with a new title copy

Turning a success screen into a toast

Refactoring the web selector component

CONTEXT

I joined Bitso right after the app rebrand, but the sign-up flow only got a cosmetic update to match this new visual identity, the experience itself hasn’t changed.

Long sign-up flows are common in financial apps. But Bitso's registration wasn't just long, it had a lot of friction. And once a user drops off, they rarely come back.

THE PROBLEM

Around 43% of users dropped off during the sign-up flow

Manual steps, avoidable errors and extra screens slowed users down

Any flow change required twice the effort, because it was still in native code

THE GOAL

We had one simple objective: increase the app and web sign-up conversion rates.

OUTCOMES

+2,000 new users per month

after only improving the country selection list

+2.47% in conversion

on the flow after adding the email autocomplete

WHAT I DID

Mapped the flow and ran a heuristic evaluation

I rebuilt the sign-up journey in Figma, from country selection to the success screen. Then I evaluated each screen with standard usability heuristics and my own product judgment.

Turned findings into hypotheses, not opinions

Every item became a testable experiment, traceable to a documented problem, and marked with its severity. The audit told me what was broken, and the funnel data helped the team decide what to fix first.

Created the testing backlog, together with the PM

I presented the audit and the opportunities to the PM and worked with him to prioritize what to test first. The backlog was adopted instead of shelved.

LAUNCHED EXPERIMENTS IN DETAIL

Removing countries Bitso doesn't operate in from the list

Even though people couldn't download the app from some countries, they could still be selected in the country of residence question, and clicking on an inoperative country opened an error. Not the kind of experience we wanted our clients to start with.

In the end, we reduced the list and I added a radio button next to the country name, making it easier to understand that it is clickable.

This increased sign-up conversion by 2.8%, which meant 2,000 more customers per month!

TEAMMATE RECOGNITION

Product Director

Small changes, BIG impact!

Conversion improvements here compound across the entire funnel.

Great work team!

Small changes, BIG impact!

Conversion improvements here compound across the entire funnel.

Great work team!

Email domain autocomplete

Email was the biggest leak in the flow. About 5% of users dropped at the email field and 17% at the verification step that followed.

Users typed their address by hand, and over 8,600 times a year the verification code never reached them because of typos.

We tested adding suggestions for the most common domains while users typed, so they only had to pick one, and sign-up conversion rose 2.47%.

IN RETROSPECT

This project reinforced a core belief of mine: the most impactful changes are the ones that make users’ lives easier, like reducing friction and offering shortcuts.

Rather than redesigning the entire flow, I focused on the smallest changes that could drive the biggest impact. This let us ship and see the results quickly.

It also taught me that when the team roadmap is not yet defined, the best way to achieve its growth goals is by working with what we already have, instead of just adding new things to the product.

This project reinforced a core belief of mine: the most impactful changes are the ones that make users’ lives easier, like reducing friction and offering shortcuts.

Rather than redesigning the entire flow, I focused on the smallest changes that could drive the biggest impact. This let us ship and see the results quickly.

It also taught me that when the team roadmap is not yet defined, the best way to achieve its growth goals is by working with what we already have, instead of just adding new things to the product.

Google sign-up button for the app

Google sign-up button for web

Exploring how to nudge users to complete the ID verification after finishing the sign-up

NEXT CASE

Scaling Nomad's referral program with continuous A/B testing

Available to work

Let’s create something great together

Chat with Maria

Let's connect

Increasing sign-up conversion at Bitso, one experiment at a time

Increasing sign-up conversion at Bitso, one experiment at a time

Increasing sign-up conversion at Bitso, one experiment at a time

Around 43% of users dropped off during the sign-up flow

Around 43% of users dropped off during the sign-up flow

Manual steps, avoidable errors and extra screens slowed users down

Manual steps, avoidable errors and extra screens slowed users down

Any flow change required twice the effort, because it was still in native code

Any flow change required twice the effort, because it was still in native code

+2,000 new users per month

after only improving the country selection list

+2.47% in conversion

on the flow after adding the email autocomplete

Mapped the flow and ran a heuristic evaluation

Mapped the flow and ran a heuristic evaluation

I rebuilt the sign-up journey in Figma, from country selection to the success screen. Then I evaluated each screen with standard usability heuristics and my own product judgment.

Turned findings into hypotheses, not opinions

Turned findings into hypotheses, not opinions

Every item became a testable experiment, traceable to a documented problem, and marked with its severity. The audit told me what was broken, and the funnel data helped the team decide what to fix first.

Created the testing backlog, together with the PM

Created the testing backlog, together with the PM

I presented the audit and the opportunities to the PM and worked with him to prioritize what to test first. The backlog was adopted instead of shelved.

Removing countries Bitso doesn't operate in from the list

Removing countries Bitso doesn't operate in from the list

Product Director

Product Director

Email domain autocomplete

Email domain autocomplete

Email was the biggest leak in the flow. About 5% of users dropped at the email field and 17% at the verification step that followed. Users typed their address by hand, and over 8,600 times a year the verification code never reached them because of typos.

We tested adding suggestions for the most common domains while users typed, so they only had to pick one, and sign-up conversion rose 2.47%.

Scaling Nomad's referral program with continuous A/B testing

Scaling Nomad's referral program with continuous A/B testing

Scaling Nomad's referral program with continuous A/B testing

Let’s create something great together

Let’s create something great together

Email was the biggest leak in the flow. About 5% of users dropped at the email field and 17% at the verification step that followed.

Users typed their address by hand, and over 8,600 times a year the verification code never reached them because of typos.