70% E2E test coverage — up from 20%
Test Strategy · Financial Services · Automation
How spriteCloud turned a broken QA process into a risk-based quality system and enabled one of Europe’s oldest impact investors to release software monthly instead of every 3–6 months.
Oikocredit is one of the world’s oldest and largest cooperative impact investors. Founded in 1975, it channels ethical investment from individuals across Europe into microfinance, agriculture, renewable energy, and community infrastructure across the Global South.
The cooperative manages over EUR 1 billion in total assets, connecting 45,200+ investors across 11 European countries with 474 partner organisations in 33 countries. 88% of end-beneficiaries are women. Over EUR 5.8 billion has been disbursed since inception.
This is not abstract finance. When the technology fails, real investors cannot access their accounts. Real communities lose access to the capital that funds their livelihoods. Quality here is not a checkbox. It is a social obligation.
The platform was in the middle of its largest structural transformation in 50 years, consolidating 13 legal entities, 6 products, and 6 countries into a single unified platform. Country by country, investor accounts needed to be migrated, tested, and transitioned.
The first country rollout, Germany, had already gone badly. Marketing was simultaneously ramping up campaigns to drive new investor registrations, while the registration system itself was broken and causing customers to drop out mid-process.
spriteCloud was hired as extra test capacity. What we found when we walked in was far deeper than an understaffed test team.
The QA Health Check was not in the SOW. Anko Beijleveld ran it on his own initiative before starting any formal test work. What it uncovered changed the entire engagement scope.
spriteCloud was hired as extra test capacity. But an experienced test manager does not start typing. They look first. What are we actually testing? Where are defects surfacing? What does the development process look like upstream?
The answer was that the problem was not the testing. It was the entire development process. User stories did not reflect how the business actually operated. Requirements had no traceability. Testers had no idea which business need a story addressed. None of this was visible until UAT.
The health check reframed the entire engagement. Not “add more testers.” Rebuild the approach.
Testing everything equally is the same as having no strategy. spriteCloud anchored the entire test approach on investor onboarding, the flow that carried the highest financial and reputational risk.
Investor onboarding was non-negotiable. If registration fails, investors cannot access their accounts. If KYC fails, the platform violates regulatory requirements. If country-specific questions appear in the wrong jurisdiction, there are legal consequences. This is where the risk lived, so this is where the testing anchored.
Getting buy-in required reframing. Business analysts had controlled requirements for years and saw QA as an intrusion. The argument: “You need 10–20 extra testers to test everything at full depth. Or you can do risk-based testing and make sure what matters most is covered properly with the team you have.” That argument landed.
Country-specific rules were pre-confirmed with business upfront and built directly into the test strategy. Entry criteria moved from “development is done” to “requirements are confirmed with business before development begins.”
Salesforce is notoriously difficult to automate. The existing team believed it could not be done with proper frameworks. They were wrong, and a formal tool evaluation proved it.
spriteCloud evaluated 4 candidates: TARA (existing record-and-playback, 1 year to build, 3 reruns needed), Playwright, Selenium, and Provar. Playwright was selected: rebuilt the full suite in 2 weeks, 95%+ stable runs, massive community, shadow DOM support for Salesforce.
The automation suite covered the complete end-to-end path: website registration → Salesforce intake → ITIN registration → confirmation back to Salesforce → data warehouse sync → customer portal access codes → first investment transaction.
External services (KYC risk analysis, face recognition) had no test environment. spriteCloud used network interception: the call went out, was intercepted, and a simulated correct response came back. Full E2E testing with no third-party dependency blocking it.
Bringing QA into refinement sessions was the hardest organisational change, and the one with the highest return. It did not just catch bugs earlier. It changed the culture.
Before, testers only saw work after user stories were written and already with developers. Beautiful process documents existed, but nobody mapped requirements to user stories. The connection was lost in translation. End users would arrive at UAT and say “but that’s not how we do it.” Then arguments. Then rework. Then delays.
When QA joined refinement sessions, the dynamic changed. Defects were filed against requirements quality. After shift left, critical bugs in UAT and production dropped to zero.
We were deep in an international HR rollout. Top management had decided all payslips would be standardised in English. One tester raised her hand in a refinement session: “What happens if an employee does not speak English? Is there a legal requirement we’re missing?”
The confirmation came back: in multiple countries on the rollout, payslips are legally required to be in the national language. If that question had not been asked, we would have built, tested, and deployed a payroll system non-compliant with national employment law across multiple countries.
That is what shift-left means in practice. One tester, one question, on a Tuesday afternoon, before a single line of code was written.
Before spriteCloud, the go/no-go question was answered with gut feeling. After: with a dashboard that stakeholders were reading before the morning news.
spriteCloud built a quality KPI dashboard tracking test case status by risk level (high/medium/low), outstanding bugs by severity, and post-UAT end-user feedback scores (target: 8–10 out of 10). The primary exit criterion: all high-risk test cases must be fully executed and successful before a release can proceed. No exceptions.
Automated regression results were embedded in the dashboard so stakeholders could see that previous releases remained stable, with no regression from sprint 6 breaking in sprint 12.
The first release was delayed by two weeks, based on dashboard data, not politics. That was the moment that changed the culture. Stakeholders started checking the dashboard first thing every morning. The go/no-go question went from an uncomfortable conversation to a 90-second review.
In early 2025, the platform launched a fully digital investor onboarding pathway, replacing 26 country websites with a single unified platform. The system spriteCloud tested is now in production.
They rewrote an automation suite in two weeks that had taken our previous team nearly a year to build. Speed like that comes from knowing exactly what you’re doing and why.
QA Lead, Oikocredit
Every team has a regression problem. Let’s fix yours in 8 weeks.
Start the conversationSee more case studies