Sprint 1: Scope it against reality
Your initiative: _______________________ Group: __________
You are the delivery lead. The money is committed. Before you can plan anything, you have to scope it against reality, and the only people who know the reality are down the hall. Interview Priya (Data) and Marcus (CIO) and use what they tell you.
Not on the table today: whether this initiative is worth the investment. That call is made: the board funded it. This is delivery scoping (what ships first, what data it needs, what “good enough” means), not business-case scoping. If your group finds itself arguing about ROI or whether to do it at all, you’ve drifted into a different conversation.
Tip: Marcus will push speed and ambition. Priya will give you the honest, data-grounded timeline. Your scope is where you reconcile the two, and you’ll have to defend it later.
1. Scoped objective (one sentence)
What, specifically, will be delivered, and what is deliberately out of scope for this first release?
- In scope: ___________________________________________
- Out of scope (for now): _______________________________
2. Data requirements (from your Priya interview)
| What data the project needs | Does RetailFlow have it, and is it good enough? | Work required first |
|---|---|---|
3. Apply the trust tool
Split the work into parts. For each, is it Average (AI produces a good-enough answer) or Precise/high-stakes (a human must stay in charge)?
| Part of the work | Average → automate | Precise/high-stakes → keep a human |
|---|---|---|
4. Define “good enough”
AI output is non-deterministic; there’s no spec to pass. So you must define fitness for purpose before deployment.
- The single metric that says “safe to go live”: __________________
- The threshold on that metric: ______________
- The maximum acceptable error rate, and what kind of error is unacceptable: ______________
Leave Sprint 1 with: a scoped objective, data requirements, a trust-tool split, and a defined “good enough.” You’ll build the rest of the plan on these.