Scalable Quality Across 18 International Banking Projects

Since 2022, the largest bank in Latin America has reduced production defects by 50%, strengthened the quality of its systems, and gained greater control over its deliveries, thanks to the integration of continuous QA with Abstracta across 18 key projects through a shift-left testing approach.

Table of contents

Meet Our Client

Our client is Latin America’s largest bank, backed by over a century of experience. It focuses on retail banking, SMEs, and corporate clients. This financial institution promotes a responsible business model that integrates environmental, social, and governance criteria to create positive and sustainable impact on society.

Presence in 18 countries
Around 70 million customers
A network with thousands of physical service points and nearly 50,000 ATMs
A global team of 100,000 people

Understanding Challenges

In 2022, bank leaders approached Abstracta with the goal of finding a quality partner to help deliver high-quality software systems for their customers.

Initially, they aimed to validate their systems through functional testing from early stages and to strengthen regression testing through automation, in order to scale test coverage across multiple projects.

The main challenge was consolidating a cross-cutting QA practice capable of sustaining quality in an ecosystem with multiple teams, technologies, and delivery cadences. Through a deep understanding of the business, over time we proposed new lines of work in observability, automation, and the use of AI with custom agents for different tasks, with the goal of driving more robust, scalable, and sustainable results.

Illustration of a person at a laptop placing a chess piece on the screen and holding a connected-nodes icon, representing strategy and thoughtful decision-making.

Solutions and Activities

From the outset, we focused on building a solid and scalable QA practice, embedded within teams and guided by a long-term vision. To address the challenges, we deployed a set of key actions:

QA discipline standardization: We defined a shift-left testing strategy and established metrics in Jira to support data-driven decision-making and enable delivered products to generate business value and customer satisfaction. We collaborated on writing user stories, introduced bug tagging based on business impact, and systematically recorded improvement opportunities.
Agile governance and team integration: We actively participated in all Scrum ceremonies using Jira and Xray. We established weekly internal meetings, monthly retrospectives, and a Meet Mentors program with senior QAs to guide the team. This helped us provide consistent, uninterrupted service quality.
Knowledge management: We centralized testing documentation in a shared, client-owned repository, including guidelines, flows, manual mind maps, and agreements to standardize the testing process. This enabled knowledge transfer and maintained traceability.
Cross-project coverage in critical initiatives: We distributed our team across 18 business areas covering retail, wholesale and corporate banking, benefits, fraud prevention, risk, operations, and optimization engines. We adopted a flexible approach, implemented testing best practices, and standardized processes while accounting for each project’s specific context.
Daily QA practice: We executed functional, exploratory, regression, integration, and mobile testing. We analyzed and designed test cases, managed users with different roles, and maintained traceability of results.
Communication: We maintained direct communication with all areas from the beginning to align expectations and delivery goals. We logged and followed up on incidents and improvements, and documented testing results to provide a clear view of system status.
Innovation: We proposed new ways to strengthen QA, including AS/400 automation, regression automation initiatives, and the use of AI agents for user story creation and test data generation, among others. We also suggested incorporating advanced metrics such as DORA and Apdex in Datadog, aligned with DevOps practices, to strengthen observability and continuous improvement.

This set of activities allowed us to move from functional testing support to an integrated shift-left testing strategy with continuous tracking throughout the development lifecycle, adapted to the client’s dynamics and capable of supporting diverse and complex projects.

Results and Achievements

Our QA contributions helped the bank make go-live decisions based on stricter criteria and data-backed confidence. The proposed strategy reduced defects, accelerated deliveries, and supported prioritization based on business value.

Quantitative results:

50% fewer production bugs, thanks to more effective early-stage validations.
Defects per user story reduced from 7 to 3 per 10 user stories, with greater clarity around criteria and business context.
Sustained QA scalability: we tripled the number of covered projects (from 6 to 18) without compromising quality.

Qualitative and management results:

More informed go-live decisions.
More precise acceptance criteria in vendor deliveries.
Coordinated functional regression across 8 critical projects in the mobile app.
Business-focused fixes through bug categorization and Jira metrics for analysis and prioritization of functionalities requiring more attention.
Accessible, centralized knowledge, with documentation, manuals, flows, reports, and working methodologies stored in a client-controlled space.

Impact on digital transformation

Integration with Apple Pay and Google Pay.
Personalized offers for different customer segments.
Development of a corporate API that streamlines financial operations.
Cloud migration of internal systems to improve efficiency and scalability.

Looking ahead, we plan to continue expanding the scope of software testing to strengthen the bank’s system quality through different lines of work: test automation, including AS/400, secure use of AI agents, and increased focus on performance testing, usability, observability, and DevOps practices.