Rebuilding a Payments Test Team in Two Quarters

Start from the work, not from a generic competency list. Build a matrix of required competencies against current proven level (0 absent, 1 aware, 2 practised, 3 able to lead) for: test automation engineering, performance test design and analysis, interface and protocol testing (ISO 8583, message-level and end-to-end), SQL and data validation, risk-based test analysis, and domain knowledge of card acquiring.

Likely result: domain knowledge and SQL are strong (level 2-3); automation engineering is at level 0 after the resignation; performance is level 1 at best; ISO 8583 is absent.

Sourcing decisions and reasoning:

• Hire one senior automation engineer/lead as a permanent role. This is the capability that must persist across releases; a contractor would leave and take the framework knowledge with them. It is also the hardest gap to close by training within two quarters.
• Hire one integration tester with proven protocol experience. ISO 8583 message-level testing is a specialist skill and the learning curve on live payment traffic is too steep for on-the-job training before this release.
• Buy as a service the first performance test cycle (load model, script development, one execution and analysis) with an explicit knowledge-transfer clause and shadowing by two permanent staff. Performance demand is spiky and the tooling is expensive; owning it fully in quarter one is poor value. Build in-house capability to run subsequent cycles.
• Train the two SQL-capable analysts in test automation foundations and risk-based test design, and train the whole team in the new domain area. Training cannot manufacture an automation lead in eight weeks.

Record the decision and the residual risk - for example, single-person dependency on the new automation lead - so it is managed rather than discovered.

Separate hygiene from motivation. Exit interviews point at career path and boring work, which are motivators, not pay. That shapes the response.

Identify retention risk per person. For each of the 12, record criticality (single point of knowledge), mobility (how easily they could leave), and stated motivation. Two or three people will carry disproportionate risk; act on those individuals specifically rather than with a team-wide gesture.

Create a visible technical career path. Define two ladders - technical (test analyst to automation specialist to test architect) and managerial - with concrete criteria. Give the strongest analysts owned areas: one owns the automation framework roadmap, one owns performance test design. Ownership is the antidote to 'boring work'.

Protect the training spend without hostage-taking. Rather than bonding people, sequence training so that it is tied to an internal role they are moving into, and distribute knowledge deliberately: pair on automation, require framework documentation, and rotate who runs the regression suite. This reduces single points of failure whether or not someone leaves.

Use the contractors deliberately. Before their eight weeks end, have them write down and hand over the test assets they hold and pair with permanent staff. Do not let critical knowledge exit with the contract.

Measure the intervention. Track regretted attrition, internal moves into the new roles, and the number of areas with a named backup. If regretted attrition stays high after two quarters, the cause is likely hygiene (workload, environment chaos, recognition) and needs a different remedy.

Quarter one - stabilise and start, do not transform. Weeks 1-2: complete the skills matrix, confirm the managed-service contract for performance with a shadowing requirement, and appoint the automation hire's interview panel with an offer target in week 4. Weeks 2-8: the two SQL analysts start automation foundations training and immediately apply it to the lowest-risk regression area, so learning produces visible output. Contractors hand over assets. Introduce a shared severity scale and defect triage discipline - cheap, immediate quality gains that build credibility. Do not attempt to automate the 900-case regression suite; pick the 80-120 highest-risk cases.

Quarter two - scale what worked. The automation lead builds the framework with the two trained analysts; the automated regression set reaches a stable, trusted core that runs unattended on each build. The new integration tester owns the ISO 8583 interface test design with supplier input. Performance capability moves in-house for the second cycle, with the managed service in an advisory role only.

Explicitly not doing: no big-bang tool migration, no attempt to make every tester an automation engineer, and no new test process layered on top of a release in execution. Capacity is the constraint - training days come out of the same 12 people who are testing the release, so schedule them between execution cycles and account for the lost capacity in the plan rather than pretending it is free.