Lesson 2 of 5
Iteration goals
11 min
Listen to this lesson
Each conversion cycle is an iteration. If its goal is vague ("convert CMQ"), it will be judged by whether SUM finished, and the program will learn little. If its goal is specific and measurable, each cycle retires a named set of risks and the program arrives at the production weekend with evidence for everything. This lesson sets goals for Contoso's cycles from the first sandbox to production.
Goals by cycle
| Cycle | Primary goal | Measurable exit |
|---|---|---|
| Sandbox (copy of EP1) | Find technical blockers and measure volumes | SUM completes; blocker list owned; first runtimes and SPDD/SPAU volumes recorded |
| CMD (development) | Produce the adaptation and Customizing transports | Readiness ATC has no open in-scope errors; finance migration Customizing in transports; SPAU complete |
| CMQ (quality) | Prove processes on converted data | Regression and integration cycles complete; defect thresholds met; one period-end close passed |
| Mock 1 | Run the runbook end to end and find gaps | Every step timed; overrun and waits analyzed; reconciliation method proven |
| Mock 2 | Fit the window and measure fallback | Duration within the window; reconciliations at zero; fallback duration measured |
| Dress rehearsal | Prove the final runbook as production | Window met with agreed contingency; sign-offs on time; runbook frozen |
| Production | Convert and release the business | Go/no-go criteria met; business released Monday 06:00 |
Contoso's iteration goals. Each exit criterion is a fact someone can check.
Notice that the goals shift from technical feasibility to process correctness to timing and repeatability. Early cycles may run slow and fail often; that is their purpose. Late cycles should hold no new kind of problem. If Mock 2 discovers a class of defect that the sandbox should have found, the program learned it too late.
Goals for each workstream in each cycle
A cycle goal becomes real when each workstream has its own part of it. For CMQ, Kenji's goal is that every adapted object reached by the regression set passes; Dan's is that reconciliation reaches zero and a close runs; Priya's is that regression and integration cycles complete within defect thresholds; Sam's is that follow-up time falls below a target because steps are now transported or scripted. Grace's PMO collects these into one cycle scorecard.
- Set the goalOne primary goal per cycle, plus workstream goals with measures
- Run the cycleConvert, follow up, test, reconcile, time everything
- Score itEach measure met, partly met or missed, with evidence
- LearnCauses of misses fixed at source; runbook and plan updated
- GateDecide whether the next cycle can start (lesson 3)
Worked example: how the rehearsals converged
Contoso’s cutover rehearsals show iteration goals doing their job. Mock 1 took 71 hours against the 60-hour window: SUM technical downtime ran 29 hours and post-processing 17. Its goal was to run the runbook end to end and find gaps, and it did: Sam tuned SUM parallelization, Dan’s team moved enrichment fixes into ECC before the weekend, and reconciliation reports were run in parallel by company code. Mock 2 took 58.5 hours (SUM 24, post-processing 11), leaving 1.5 hours of float, and measured the fallback at 9 hours. It met its goal of fitting the window, but the float was below the steering group’s 4-hour contingency minimum, so the dress rehearsal’s goal included recovering contingency. The dress rehearsal finished with 4 hours of contingency and the runbook was frozen.
- One primary goal per cycle, written as the risk it retires.
- Three to six measures, each with a target, an owner and an evidence source.
- A scoring rule agreed in advance: met, partly met or missed, and what partly met means.
- Carry-forward: every missed measure becomes a named goal or entry condition for the next cycle.
Using the Wave 2 horizon
Iteration goals should also protect what comes after go-live. Contoso plans to load Alternadores del Norte into the converted system as company code 1910 soon after the conversion. The data team needs a stable converted quality system to start its mock loads. So CMQ's goals include a target that matters to Wave 2: the converted system leaves room to configure Mexican company code 1910 (MXN) and plant 1910 Monterrey, and it can be refreshed for Wave 2 mock loads after go-live without re-running conversion fixes.
What would you do?
Sam reports that the sandbox conversion 'went well': SUM finished in 40 hours with no stops. Nobody ran the finance reconciliation, and SPAU was not started. Grace has to decide whether the sandbox cycle met its goal. What is the right call?
- Give every conversion cycle a primary goal and measurable exit criteria.
- Goals shift from technical feasibility to process correctness to timing and repeatability.
- Each workstream owns measures in each cycle; the PMO scores them on one scorecard.
- A new class of problem in a late cycle is a planning signal; protect follow-on work such as Wave 2.
Check your understanding · not graded
Lesson check
1. What is the primary goal of Contoso's Mock 2?
2. How should iteration goals change from early to late cycles?
3. A new class of problem appears in Mock 2. What does it signal?
4. Which is a measurable exit criterion for CMD?
5. Mock 2 took 58.5 hours with 1.5 hours of float; the steering minimum is 4 hours of contingency. What belongs in the dress rehearsal’s goal?