Lesson 6 of 6
Maturity roadmap
12 min
No TCoE starts mature. The question is how to grow it deliberately over 12 to 24 months so that each step pays for the next. A roadmap ties capability growth to the calendar of change you built earlier, so investments land before the upgrades that need them.
Four capability tracks
| Track | Starting point (typical after go-live) | Target in 24 months |
|---|---|---|
| Regression pack | Project test cases, partly outdated, not risk-ranked | Maintained, risk-ranked pack per critical process, reviewed each release |
| Automation | Few or no scripts; or project scripts that no longer run | Majority of stable regression automated, with low false-failure rate and clear ownership |
| Test data | Manual data creation for each cycle | Data recipes per process, refresh and masking routine, self-service requests |
| Process and governance | Ad hoc requests and project-specific standards | Service catalog, intake, standards, metrics and a quality council |
Four tracks to mature in parallel. Progress on one often depends on another.
- Months 0–3FoundationService catalog, intake, standards; inventory and risk-rank the regression pack; first metrics baseline.
- Months 3–6Stabilize regressionRefresh test cases for critical processes; data recipes for order to cash and procure to pay; first upgrade run on the calendar.
- Months 6–12Automate the coreAutomate stable, high-risk, high-frequency processes; measure cost per cycle; business test network formalized.
- Months 12–18ScaleExtend automation and data recipes; release assessment as a routine service; self-service data requests.
- Months 18–24OptimizeRisk-based regression selection per release; trend metrics drive investment; continuous improvement cycle.
Choosing what to automate
Automation pays when a test is run often, the process is stable and the business risk is high. It costs when the process changes every release, the data is hard to set up, or the screens are volatile. Use a simple scoring of frequency, stability, risk and data effort, and automate from the top of the list. For Public Edition, start with what SAP provides (Post-Upgrade Tests and the standard automates of the Test Automation Tool), then maintain customer automates for customized processes. For Private Edition, choose tooling against your standards, keeping the regression design tool-neutral so scripts can be replaced without losing the test logic.
Test data as a capability
Most automation programs stall on data, not scripts. A test that needs a customer with credit limit, a material with stock in storage location 171A and valid pricing will fail if any of those are missing or consumed by the last run. Build data recipes: documented, repeatable ways to create or select the data each process test needs, ideally automated themselves. Pair them with a refresh and masking routine for test systems, agreed with security and data protection, so production-like data is available without exposing personal data.
Funding the roadmap
Fund the TCoE in two parts: a run budget for core services (regression ownership, data, environments, reporting) and investment for roadmap items (automation builds, data tooling), justified by the reduction in cost per cycle and the upgrade windows they protect. Tie investment releases to milestones so leadership sees each step pay off before funding the next.
What would you do?
Contoso's TCoE has budget for one major investment this year. The options are: automate the full plan-to-produce process (which changes every quarter), build data recipes and automation for order to cash (stable, high volume, critical), or buy a second tool for a team that prefers it. What should the TCoE lead recommend?
- Grow four tracks in parallel: regression pack, automation, test data, and process and governance.
- Phase the roadmap over 12–24 months and land capabilities before the upgrades that need them.
- Automate stable, frequent, high-risk processes first; keep test design tool-neutral.
- Treat test data as a capability, and fund the TCoE as run budget plus milestone-based investment.
Check your understanding · not graded
Lesson check
1. Which process is the best first candidate for automation?
2. Why do many automation programs stall?
3. For Public Edition, where should automation start?
4. How should TCoE investment be funded?