Case Study
A UK mid-tier financial services firm with a 23-person QA team adopted Softomate's Testomate platform to automate regression and UAT testing on its customer-facing platform, cutting full UAT cycle time from 14 days to 3, improving defect-detection rate by 37%, and freeing the QA team to focus on exploratory and risk-based testing rather than repetitive scripted runs.
13 min read By Deen Dayal Yadav, Founder & AI Automation Director
UK mid-tier financial services firm, 23-person QA team, mission-critical customer-facing platform
A UK mid-tier financial services firm with a 23-person QA team adopted Softomate's Testomate platform to automate regression and UAT testing on its customer-facing platform, cutting full UAT cycle time from 14 days to 3, improving defect-detection rate by 37%, and freeing the QA team to focus on exploratory and risk-based testing rather than repetitive scripted runs.
A UK mid-tier financial services firm operating a customer-facing platform serving over 400,000 retail customers had a 23-person QA function responsible for testing every release of its core platform. The platform was business-critical: outages or material defects had direct customer impact and regulatory implications under the firm's FCA permissions. Release cadence was monthly, with each release preceded by a structured User Acceptance Testing cycle that the firm's risk-and-compliance function required as a sign-off gate before any release reached production.
The UAT cycle was the operational pinch point. Full UAT covered approximately 1,400 scripted test cases across the platform's seven core modules (customer onboarding, account management, payments, transfers, document management, notifications, and reporting). The cycle was executed largely manually by the QA team, with each test case taking on average 4 minutes to execute including the time to set up test data, walk through the scripted steps, verify the expected outcome, and log the result. Across 1,400 test cases at 4 minutes each, the raw execution time alone was over 93 hours of QA effort per cycle. With overlap, parallelism, and the inevitable defect investigation work that surfaced during testing, the actual elapsed UAT cycle was averaging 14 working days.
The 14-day cycle was the binding constraint on release cadence. The firm wanted to move to a fortnightly release cycle to reduce the risk window between feature development and production, but with UAT consuming 14 working days of each cycle, the maths did not allow it. The QA director had run the analysis twice in the previous year and concluded that fortnightly releases would require either a 50% increase in QA headcount (a roughly £430,000 annual cost increase) or a structural change in how UAT was executed.
The defect-detection profile added the second pressure. The QA director's analysis of production incidents over the previous twelve months showed that approximately 38% of defects that reached production were defects that the existing scripted UAT process would not have caught regardless of cycle time, because they were edge cases, race conditions, or interaction effects between modules that scripted testing was not designed to expose. The QA team's most experienced testers had repeatedly raised this in retrospectives but the team simply did not have the time to do meaningful exploratory or risk-based testing because the scripted regression cycle consumed almost their entire bandwidth.
The team-morale dimension was the third concern. The QA director had run an internal engagement survey six months prior to the engagement that showed concerning patterns. The team's most experienced testers (those with five-plus years of experience in financial services testing) reported the lowest engagement scores, primarily because their working day was dominated by repetitive scripted execution that did not use their judgment or skill. Three senior testers had given notice or were close to giving notice during the previous twelve months, and replacing them was both expensive and time-consuming given the scarcity of experienced financial services QA professionals in the London market.
The regulatory dimension was the fourth concern. As an FCA-regulated firm, the company's release process was subject to formal regulatory scrutiny, including the firm's Senior Manager Certification Regime obligations on the chief operating officer. Any change to how UAT was executed needed to satisfy not only internal quality standards but the documented expectations of the firm's regulator. The QA director was acutely aware that introducing automation into the UAT process would need to be supported by clear evidence that the automated outputs were at least equivalent in rigor to the manual outputs they replaced.
The platform context added a fifth dimension. The firm's customer-facing platform was a complex web application built on a modern JavaScript framework, with a substantial mobile-app channel that shared underlying APIs but had its own UI testing requirements. Any automation framework would need to cover both the web and mobile channels reliably, and would need to handle the financial-services-specific UI patterns (multi-factor authentication, payment confirmations, transaction signing flows) that off-the-shelf testing tools often struggled with.
Softomate proposed adoption of its Testomate platform, an AI-augmented test automation framework specifically designed for financial services platforms where regulatory documentation requirements add overhead to automated testing tooling. Testomate combines a Playwright-based browser automation engine with an LLM-driven test interpretation layer that allows test cases to be expressed in structured natural language rather than code, while still producing reliable automation outcomes that satisfy audit requirements.
The discovery phase mapped the firm's existing 1,400 test cases by complexity and automation suitability. Approximately 1,180 of the cases were straightforward scripted flows (login, complete a form, verify an outcome) that were natural candidates for browser-driven automation. The remaining 220 cases involved either external system integration that required specific test environment setup, or visual verification that required human judgment, and were retained as manual cases.
The platform-evaluation rationale deserves explicit note. The QA director had previously evaluated three alternative test automation platforms before approaching Softomate: a code-based Playwright implementation, a commercial low-code platform (which Softomate identified during discovery as one of the well-known names in the QA automation space), and a record-and-replay tool. The code-based approach was viable but would have required the QA team to develop programming skills the team did not currently have, with a learning curve estimated at six to nine months before useful productivity. The commercial low-code platform was capable but the per-test-execution pricing made it unviable at the firm's volume. The record-and-replay tool was operationally simple but produced brittle tests that broke on minor UI changes, which the QA director had personally experienced in a previous role. Testomate's structured natural-language approach was selected because it offered the productivity of a low-code tool without the per-execution pricing and without the brittleness of record-and-replay.
The Testomate platform implementation involved three workstreams running in parallel over a 14-week engagement. The first was test-case migration: converting the 1,180 automation-suitable scripted cases from their existing format (a combination of Word documents, Excel sheets, and a legacy test management tool) into Testomate's structured natural-language format. Softomate's team led the first 200 cases as a worked example for the QA team, who then took over the bulk of the migration with Softomate review and refinement.
The second workstream was test environment configuration. Testomate needed reliable access to a representative test environment with realistic test data that could be reset to known state between test runs. The firm's existing test environment had been built incrementally over years and had several known issues around data consistency. Softomate worked with the firm's infrastructure team to rebuild the test environment with infrastructure-as-code, with database state reset built into the test orchestration layer.
The third workstream was the integration between Testomate and the firm's existing release management toolchain. Test results from Testomate runs needed to flow into the firm's existing release gate dashboards in a format that satisfied the risk-and-compliance function's audit requirements. Softomate built a custom adapter that produced the firm's required reporting format directly from Testomate's structured test results, eliminating the manual reporting step that had previously consumed significant QA time at the end of each cycle.
The mobile-app coverage was a fourth workstream added during week three of the engagement after the QA director identified the gap. Testomate's mobile testing capabilities were extended for this engagement to cover the firm's iOS and Android apps via Appium under the hood, with the same structured natural-language test format. This allowed the firm's mobile-specific test cases (approximately 240 of them) to be automated alongside the web cases under a single test management approach, rather than maintained as a separate parallel suite.
The change-management workstream was treated as central to the engagement rather than as a peripheral concern. The QA team's most experienced testers were involved in the test-case migration from the outset, with Softomate explicitly framing the migration as a transition from script-execution to test-engineering: the testers were learning to design and maintain a test framework, not just execute tests within it. The framing turned out to be operationally important. The senior testers, who had been the most disengaged before the engagement, became the most engaged during and after it because the work used their judgment and produced visible outputs that they could be proud of.
The audit and regulatory dimension received explicit engineering attention. Every Testomate test run produced a complete structured evidence package: the specific test cases executed, the inputs used, the actual outputs observed, the pass-or-fail outcome, and a full execution trace including screenshots at key verification points. The package was structured to satisfy the firm's internal compliance documentation requirements and was reviewable by the firm's external audit function. Softomate also engineered the system to ensure that no test case could be modified during an active test run, preserving the integrity of the audit trail.
Full UAT cycle time fell from 14 working days to 3 working days, an 11-day compression. The 1,180 automated test cases executed in approximately 6 hours of unattended wall-clock time, with defect investigation and the 220 retained manual cases consuming the remainder of the 3-day cycle. The QA team's calendar moved from being almost entirely consumed by scripted execution to having significant capacity for exploratory and risk-based testing.
The fortnightly release cadence the QA director had wanted became feasible within three months of Testomate launch. The firm moved to fortnightly releases starting in the fourth month post-launch, with no increase in QA headcount and no degradation in defect-detection quality.
The defect-detection profile improved measurably. Production defect rate per release fell by approximately 37% in the first six months post-launch, despite the doubled release cadence. The QA director attributed the improvement primarily to the team's newly available time for exploratory and risk-based testing, which surfaced the kinds of edge cases that scripted testing had structurally been unable to catch. The firm's QA function transitioned from being primarily a scripted-execution function to being primarily a quality-engineering function.
The regression run time reduction was the most operationally visible change. Specific regression suites that had previously taken multiple hours to execute manually now ran in minutes via Testomate. The team began running targeted regression suites in response to specific code changes during development rather than only during the formal UAT cycle, which surfaced defects significantly earlier in the development cycle and reduced the cost of fixing them.
The audit-trail benefit was the outcome the risk-and-compliance function valued most. Every Testomate test run produced a complete structured evidence package suitable for FCA audit purposes, including the specific test cases executed, the inputs used, the actual outputs observed, and the pass-or-fail outcome. Producing this evidence had previously required manual gathering at the end of each release cycle; with Testomate, it was a byproduct of the test execution itself.
The team-morale outcome was substantial and durable. The senior testers' engagement scores in the firm's next internal engagement survey moved from the lowest in the firm to the highest in the company-operations function. The three senior testers who had been close to giving notice all chose to stay; one was subsequently promoted into a test-engineering lead role created specifically to recognise the new senior-judgment requirements of the team's work. The QA director described this single outcome as the one she had been least confident the engagement could achieve and the one she ultimately valued most.
The mobile-app coverage extension proved disproportionately valuable. The firm's mobile channel had previously been the source of approximately 31% of production defects despite representing roughly 45% of customer transaction volume, a disproportion the QA director had been unable to address with the available QA resourcing. With mobile testing automated under Testomate, mobile-specific defect rate fell by 52% in the first six months post-launch, closing the gap with the web channel's defect rate to within acceptable tolerance.
The avoided headcount cost was the clearest financial benefit. The firm had been on the cusp of approving a 50% expansion of the QA function to handle the move to fortnightly releases; Testomate made that hire unnecessary, representing approximately £430,000 of annual avoided cost. Softomate's engagement cost was recovered within 5 months of go-live.
The firm has since extended Testomate to cover the testing of a new mobile-app channel that launched in the following quarter, and has begun discussions with Softomate about extending Testomate into security regression testing for the platform's authentication and authorisation flows.
Three things, with the benefit of hindsight, would have made the engagement smoother. First, the test-case migration was operationally larger than the discovery phase estimated; the firm's 1,180 automation-suitable test cases had inconsistent formatting across the three different source formats, which required substantial cleanup before they could be migrated. Future engagements should treat the source format consolidation as an explicit pre-migration workstream. Second, the test environment rebuild was correctly anticipated as a major workstream but was initially planned to happen in parallel with the test-case migration; in practice, the migration repeatedly hit environment-stability issues that delayed the migration team's progress, and the engagement would have run more smoothly with the environment stabilisation completed before the migration began. Third, the change-management framing (test-engineering vs script-execution) was identified mid-engagement rather than as a planning input; subsequent engagements have made this framing explicit from the first stakeholder conversation, which produces visibly smoother team adoption.
The strategic outcome that surfaced gradually over the months following deployment was the firm's changed relationship with release frequency. Before the engagement, releases were treated as risk events: each release was a moment of elevated production-incident probability, and the team's preference was for fewer, larger releases to spread the risk-event count. After the engagement, releases became routine: each release carried smaller change scope, smaller risk surface area, and a structurally lower probability of incident. The chief operating officer reported in a board update that the firm's production-incident rate per release had fallen by 71% across the first year following Testomate launch, an outcome that compounded over time as the team's release confidence increased and the development cycle compressed in response.
The next phase of work under discussion focuses on extending Testomate from functional and regression testing into performance and security testing, both of which currently require specialist tooling and specialist skills that the firm's QA team does not have in-house. The hypothesis is that the structured natural-language approach can be extended to these testing domains with similar productivity benefits, though Softomate has been clear that the architectural complexity of performance and security testing requires careful design rather than direct extension of the existing approach.
The first-year retrospective captured several outcomes that compounded across the year. Production defect rate per release stabilised at the post-launch reduced level across all twelve months following deployment, with no quarter showing regression toward the pre-engagement baseline. The fortnightly release cadence held throughout the year, and the firm subsequently evaluated moving to weekly releases at the end of the second year, a conversation that would have been unimaginable under the pre-engagement setup. The QA function's engagement scores in the firm's annual internal survey moved from the bottom quartile in the year before the engagement to the top quartile in the year following, with specific positive comments from team members referencing both the change in working day and the visible career-development paths the engagement had opened.
The team-retention outcome had material commercial impact. The QA function had historically had relatively high voluntary turnover in financial services (the QA director's previous role had averaged roughly 22% annual voluntary turnover, in line with London-market norms for the segment). In the year following Testomate deployment, voluntary turnover in the QA function fell to under 5%, with three team members specifically citing the change in working day as a primary factor in their decision to stay. The avoided replacement cost across the year was estimated at approximately £180,000 in recruitment fees and onboarding overhead, on top of the original avoided-headcount saving from not needing to hire additional testers for the fortnightly release cadence.
The audit relationship with the firm's external compliance function became materially smoother across the year. The structured evidence packages that Testomate produced for every release were used directly in the firm's audit submissions, with the external auditor specifically commenting on the quality and consistency of the evidence as a positive change. The firm's risk-and-compliance director described the change as one of the most operationally valuable outcomes of the engagement, comparable in significance to the visible release-cadence and defect-rate improvements.
The strategic outcome the chief operating officer valued most was the firm's repositioning in its own internal narrative about software delivery. Before the engagement, software releases were treated as risk events. After the engagement, software releases became routine. The chief operating officer described this in a board update as a cultural shift that affected the entire engineering organisation, not just the QA function: developers shipped code more confidently, product managers planned smaller more frequent feature releases, and the operations team's incident-response posture moved from anxious anticipation to routine readiness. The cumulative effect was a development organisation that delivered more value per quarter with less perceived risk than the equivalent organisation a year earlier.
The hiring funnel for the QA function changed measurably across the year. The firm's offer-acceptance rate on senior QA roles rose from the historical baseline of approximately 48% to over 75% during the year following deployment, with several accepted candidates specifically citing the firm's reputation for treating QA as a quality-engineering function rather than a script-execution function as a primary factor in their decision. The QA director reported that the firm had moved from struggling to attract experienced financial services QA professionals into being one of the few firms in the London market actively cited in the QA community as a destination for testers who wanted to do meaningful test-engineering work. The reputational shift, which had no specific marketing investment behind it, was an entirely organic outcome of the team members talking about the change in their working day to peers in the wider London QA community. The cumulative recruiting cost saving across the year, calculated against historical recruitment-fee spend on the QA function, was approximately £90,000, an entirely incidental benefit that the QA director recognised only at the year-end budget review.
Related service: Testomate and Automation Test Engineering London. Further reading: Agentic AI Workflows UK Companies and Agile Software Development Lifecycle.
Anonymised client engagement. Identifying details modified for confidentiality. Outcome ranges reflect typical results from similar projects.
Names withheld to preserve confidentiality.
Work with us
Every project we take on has a measurable outcome. Talk to our London team and we will show you exactly how we would approach your challenge.
Softomate Solutions Limited
Deen Dayal Yadav
Online