Compliance and improvement

Standards drift. Teams under pressure skip a field, partners bring their own habits, and templates age while SAP releases move on. The TCoE needs two light routines: one that checks whether standards are followed, and one that changes the standards when they are wrong. Both must cost little, or they will be the first thing dropped in a busy release.

Light conformance reviews

Do not review everything. Sample. After each release and at each project gate, a test architect from outside the team checks a small random sample against a short checklist and scores it. Contoso samples ten passed results per process per release, plus every waiver and every severity 1 or 2 defect.

Report the conformance rate per team and per check, not as a league table of people. A pattern matters more than a single miss: if one check fails across all teams, the standard or the tool is probably wrong. If one team fails many checks, it needs coaching, not a warning. sap-tst-310 showed the same sampling technique for partner evidence; the TCoE applies it to everyone, including itself.

Governing the playbook

A playbook is a product with versions. Every standard and template has an owner, a version number and a change log. Anyone can raise a change request: a project that found a gap, a tester who keeps hitting an impractical rule, an auditor. The TCoE triages requests monthly, and the quality council approves changes to the policy and practices. Templates change with TCoE approval alone, as long as they stay within the practices.

Retrospectives that change something

Every release and every project phase ends with a short retrospective. CTAL-TM v3.0 lists retrospectives as one of the ways to improve the test process, alongside model-based and analytical approaches. The TCoE adds one rule: each retrospective produces at most three actions, each with an owner and a date, and the actions go into the improvement backlog where the quality council can see them. A retrospective with fifteen actions changes nothing.

Measure the playbook itself with three numbers each quarter: the conformance rate per check (is it rising?), the number of waivers per release (are criteria realistic, or being waived routinely?), and the share of changes with a recorded test decision (is the policy being applied?). A rising waiver count on one criterion is a signal to revisit that criterion, not to tolerate more waivers.

Using a maturity model without being ruled by it

sap-tst-401 used TMMi and the IDEAL cycle to plan the TCoE's overall growth. For standards, the model is a checklist of what good practice expects, not a target in itself. Note that the TMMi Foundation released TMMi 2.0 in 2026, which renamed some process areas and added a new level 2 process area, Implementation and Habit, about making practices stick. That is exactly the problem this lesson addresses. Check the current framework document before using level names or process areas in an assessment.