The Launch Date Nobody Owns
A mid-size software firm is launching a major version in 10 weeks. The cross-functional team includes engineering (14 people), a part-time designer, one product marketer, and support. Status: core features are 80% built; two features are unstarted and were promised to a launch partner. QA has not begun regression testing. Documentation is unwritten. The product marketer has a launch plan but no finalized pricing. Support has had no training. The VP of Sales has already told the partner the date is firm. Engineering says the two unstarted features need 4 weeks each and cannot run in parallel with the same two engineers. No one owns the launch readiness checklist; the PM assumed someone else did.
1. Diagnose the launch situation. What is the actual critical path, and what is the most likely failure mode?
Read the worked answer
The critical path is not the feature work - it is the unstarted partner features plus regression testing, which are serialized by the same two engineers.
- Two unstarted features at 4 weeks each, not parallelizable = 8 weeks on the critical path.
- Regression testing cannot meaningfully start until the build is feature-complete, and it has not begun. That adds weeks after the 8.
- Ten weeks remain. The date is arithmetically infeasible if both features ship.
The most likely failure mode is a late, untested launch with an untrained support org and no pricing - the classic pattern where engineering absorbs all attention and the go-to-market readiness work (docs, training, pricing) gets compressed into the final days or skipped. The absence of a checklist owner guarantees these items surface late. The partner promise is a commitment made without the delivery team's input, which is a governance problem masquerading as a schedule problem.
2. As PM, propose a concrete recovery plan with a clear recommendation on scope and the date.
Read the worked answer
Recommendation: hold the date, cut scope, and name an owner.
- Scope decision. Ship one partner feature; defer the other with a written, dated commitment to the partner. Attempting both guarantees a late, untested launch - the worst outcome for the partner too. If the VP of Sales objects, present the arithmetic: 8 weeks of serial feature work plus regression exceeds 10 weeks.
- Protect the testing tail. Freeze features with enough runway for regression. An untested launch into a partner integration risks the relationship more than a deferred feature.
- Assign launch-readiness ownership now. One named owner (PM or product marketer) for a checklist: pricing finalized, docs drafted, support trained, rollback plan. Give each item a date and an owner; review weekly.
- Fix the governance gap. The partner commitment was made without engineering input. Institute a rule that external date promises require a delivery-team estimate first.
A launch is a system, not a build. The recovery plan has to schedule the go-to-market work with the same seriousness as the code, or the date slips anyway - just later and more expensively.