CASE 01 · DIAGNOSIS
Onboarding flow
The root cause wasn't where everyone was looking.
Qualitative research
Ops & business
Strategic decision
THE PROBLEM
The answer was with the operations team
“Every error in the onboarding flow had a double cost: a user who dropped off, and a case that operations had to resolve manually.”
The operations team regularly received errors and inconsistencies in the data users entered during the onboarding process. However, by design, we had no clarity on at which specific moments these errors occurred, what caused them, or what the real cost was of resolving them manually.
There was a gap between what the product claimed to experience and what users actually lived — and the people who knew this best weren't designers, they were operations people.
THE PROCESS
Before designing, we researched
Qualitative
Qualitative interviews
Conversations with key people from the operational team, selected through direct contact and acquainted with the flow's errors. We captured not just the visible errors in the data, but the perceptions, frustrations, and workarounds the team had developed.
5 INTERVIEWS · OPERATIONS TEAM
Synthesis
Critical points map
A cross-reference between qualitative findings from interviews and user behavior data that allowed us to identify the screens with the highest error and drop-off rate — not just a reconstruction of the complete flow, but an X-ray of its fractures.
8 CRITICAL SCREENS IDENTIFIED
🟣 VISUAL — SERVICE BLUEPRINT · ONBOARDING FLOW
THE FINDINGS
What we found — and what we decided
We prioritized the findings with the highest operational cost and the most direct impact on user conversion.
Insight /
Key Decision
01
Document capture of insufficient quality
The provider was sensitive to lighting conditions. Operations had to manually review and validate each case.
Migrate to a more robust provider against lighting variations and with better support for the region's document types.
02
Incomplete or inconsistent residence data
Imprecise data in address fields that operations had to correct or escalate.
Integrate an API with real-time autocomplete that cross-references official databases.
03
Inconsistent employment activity and income
Inconsistent information that forced operations to request income or banking certification.
Cross-validated fields and alert messages before submission.
04
Users on blacklist with no feedback
No clear information about the status of the request, generating user frustration.
Redesign the screen with clear status messages and guidance on next steps.
05
Lengthy flow with no user awareness
A long process due to regulation. Users didn't understand the importance of entering correct data.
Review the journey's hierarchy to communicate the importance of each step.
🟣 VISUAL — SCREENS WITH THE HIGHEST ERROR CONCENTRATION
THE SOLUTION
From research to production
01·BIOMETRICS
New biometric provider
We migrated to a provider with better support for the region's document types and greater robustness in facial capture, reducing cases that escalated to operations for manual review.
Implemented
02·ADDRESS
Address autocomplete API
We implemented an API that guides users in real time as they enter their residential address, reducing human error in a field that generated a significant amount of rejections and manual corrections from the operations team.
Implemented
🟣 Production prototype — biometric validation & address autofill
5
Findings identified
2
Solutions in production
The impact was qualitatively evident in the reduction of cases escalated to the operations team and in the elimination of critical blockers that prevented users from completing the flow. In a fintech where the onboarding flow is the first conversion point, reducing friction in biometric capture and address entry has a direct effect on the onboarding completion rate.
“Some of the most valuable improvements aren't born from visual redesign. They're born from listening to the people who operate the product from the inside.”
See next case →
CASE 01 · DIAGNOSIS
Onboarding flow
The root cause wasn't where everyone was looking.
Qualitative research
Ops & business
Strategic decision
THE PROBLEM
The answer was with the operations team
“Cada error en el flujo de vinculación tenía un costo doble: un usuario que abandonaba y un caso que operaciones debía resolver manualmente.”
“Every error in the onboarding flow had a double cost: a user who dropped off, and a case that operations had to resolve manually.”
The operations team regularly received errors and inconsistencies in the data users entered during the onboarding process. However, by design, we had no clarity on at which specific moments these errors occurred, what caused them, or what the real cost was of resolving them manually.
There was a gap between what the product claimed to experience and what users actually lived — and the people who knew this best weren't designers, they were operations people.
THE PROCESS
Before designing, we researched
Qualitative
Qualitative interviews
Conversations with key people from the operational team, selected through direct contact and acquainted with the flow's errors. We captured not just the visible errors in the data, but the perceptions, frustrations, and workarounds the team had developed.
5 INTERVIEWS · OPERATIONS TEAM
Synthesis
Critical points map
A cross-reference between qualitative findings from interviews and user behavior data that allowed us to identify the screens with the highest error and drop-off rate — not just a reconstruction of the complete flow, but an X-ray of its fractures.
8 CRITICAL SCREENS IDENTIFIED
🟣 VISUAL — SERVICE BLUEPRINT · ONBOARDING FLOW
THE FINDINGS
What we found — and what we decided
We prioritized the findings with the highest operational cost and the most direct impact on user conversion.
#
Insight
Key Decision
01
Document capture of insufficient quality
The provider was sensitive to lighting conditions. Operations had to manually review and validate each case.
Migrate to a more robust provider against lighting variations and with better support for the region's document types.
02
Incomplete or inconsistent residence data
Imprecise data in address fields that operations had to correct or escalate.
Integrate an API with real-time autocomplete that cross-references official databases.
03
Inconsistent employment activity and income
Inconsistent information that forced operations to request income or banking certification.
Cross-validated fields and alert messages before submission.
04
Users on blacklist with no feedback
No clear information about the status of the request, generating user frustration.
Redesign the screen with clear status messages and guidance on next steps.
05
Lengthy flow with no user awareness
A long process due to regulation. Users didn't understand the importance of entering correct data.
Review the journey's hierarchy to communicate the importance of each step.
🟣 VISUAL — SCREENS WITH THE HIGHEST ERROR CONCENTRATION
THE SOLUTION
From research to production
01·BIOMETRICS
New biometric provider
We migrated to a provider with better support for the region's document types and greater robustness in facial capture, reducing cases that escalated to operations for manual review.
Implemented
02·ADDRESS
Address autocomplete API
Se implementó una API que guía al usuario en tiempo real mientras ingresa su dirección de residencia, reduciendo el error humano en un campo que generaba una buena cantidad de rechazos y correcciones manuales por parte del equipo de operaciones.
Implemented
🟣 Production prototype — biometric validation & address autofill
5
Findings identified
2
Solutions in production
The impact was qualitatively evident in the reduction of cases escalated to the operations team and in the elimination of critical blockers that prevented users from completing the flow. In a fintech where the onboarding flow is the first conversion point, reducing friction in biometric capture and address entry has a direct effect on the onboarding completion rate.
“Some of the most valuable improvements aren't born from visual redesign. They're born from listening to the people who operate the product from the inside.”
See next case →
CASE 01 · DIAGNOSIS
Onboarding flow
The root cause wasn't where everyone was looking.
Qualitative research
Ops & business
Strategic decision
THE PROBLEM
The answer was with the operations team
“Every error in the onboarding flow had a double cost: a user who dropped off, and a case that operations had to resolve manually.”
The operations team regularly received errors and inconsistencies in the data users entered during the onboarding process. However, by design, we had no clarity on at which specific moments these errors occurred, what caused them, or what the real cost was of resolving them manually.
There was a gap between what the product claimed to experience and what users actually lived — and the people who knew this best weren't designers, they were operations people.
THE PROCESS
Before designing, we researched
Qualitative
Qualitative interviews
Conversations with key people from the operational team, selected through direct contact and acquainted with the flow's errors. We captured not just the visible errors in the data, but the perceptions, frustrations, and workarounds the team had developed.
5 INTERVIEWS · OPERATIONS TEAM
Synthesis
Critical points map
A cross-reference between qualitative findings from interviews and user behavior data that allowed us to identify the screens with the highest error and drop-off rate — not just a reconstruction of the complete flow, but an X-ray of its fractures.
8 CRITICAL SCREENS IDENTIFIED
🟣 VISUAL — SERVICE BLUEPRINT · ONBOARDING FLOW
THE FINDINGS
What we found — and what we decided
We prioritized the findings with the highest operational cost and the most direct impact on user conversion.
#
Insight
Key Decision
01
Document capture of insufficient quality
The provider was sensitive to lighting conditions. Operations had to manually review and validate each case.
Migrate to a more robust provider against lighting variations and with better support for the region's document types.
02
Incomplete or inconsistent residence data
Imprecise data in address fields that operations had to correct or escalate.
Integrate an API with real-time autocomplete that cross-references official databases.
03
Inconsistent employment activity and income
Inconsistent information that forced operations to request income or banking certification.
Cross-validated fields and alert messages before submission.
04
Users on blacklist with no feedback
No clear information about the status of the request, generating user frustration.
Redesign the screen with clear status messages and guidance on next steps.
05
Lengthy flow with no user awareness
A long process due to regulation. Users didn't understand the importance of entering correct data.
Review the journey's hierarchy to communicate the importance of each step.
🟣 VISUAL — SCREENS WITH THE HIGHEST ERROR CONCENTRATION
THE SOLUTION
From research to production
01·BIOMETRICS
New biometric provider
We migrated to a provider with better support for the region's document types and greater robustness in facial capture, reducing cases that escalated to operations for manual review.
Implemented
02·ADDRESS
Address autocomplete API
We implemented an API that guides users in real time as they enter their residential address, reducing human error in a field that generated a significant amount of rejections and manual corrections from the operations team.
Implemented
🟣 Production prototype — biometric validation & address autofill
5
Findings identified
2
Solutions in production
The impact was qualitatively evident in the reduction of cases escalated to the operations team and in the elimination of critical blockers that prevented users from completing the flow. In a fintech where the onboarding flow is the first conversion point, reducing friction in biometric capture and address entry has a direct effect on the onboarding completion rate.
“Some of the most valuable improvements aren't born from visual redesign. They're born from listening to the people who operate the product from the inside.”
See next case →