1
Original ideaBuild a pickup-slot selector
Customer outcomeCollect an order in a feasible promised window.
On-time collection82% → ≥90%Proposed; at 100 pickups,
82 → at least 90.
82 → at least 90.
Keep the destination.
Question the vehicle.
Rewrite a list of features as customer outcomes, success measures and a flexible sequence the team can discuss.
Harbor Pantry / Q1 2027Fictional demonstration · proposed outcomes
2
Original ideaAdd reorder-from-last-basket
Customer outcomeRepeat an appropriate basket with less effort.
Repeat completion60% → ≥75%Proposed; at 120 shoppers,
72 → at least 90.
72 → at least 90.
3
Original ideaAdd a staff packing checklist
Customer outcomeReceive a complete order without a support follow-up.
Missing-item complaints6% → ≤3%Proposed; at 200 orders,
12 → at most 6.
12 → at most 6.
Keep the destination flexible. Capacity, measurement definitions, permissions and effort are not yet confirmed. These are planning proposals, not ship dates or achieved improvements.
A fictional mapping, actually prepared
Original: Build a pickup-slot selector.
Enable customers to collect an order in a feasible promised window.
Success is on-time collection, not the existence of a selector.
Bring a roadmap and a reason.
Supply initiatives, target customers, company direction, known baselines and constraints. Identify uncertain dates and missing evidence.
Leave with a clearer map.
Receive original-to-outcome mappings, key success measures, dependency notes, assumptions and quarter-level release windows.
For the people explaining what comes next.
Useful for product leaders, operators and cross-functional teams who want to explain the customer or business change behind a plan. It helps review strategic intent without committing to one implementation too early.
It is not a software builder, project-system integration or proof of a future revenue gain. If the customer problem is unknown, the next step is a question rather than a made-up metric.
Use the included instructions.
Extract the ZIP and load SKILL.md using a supported host method, or have your file-capable assistant read it. Supplied-text planning needs no additional service API key. Assistant/model access and fees still apply; uploads need host file tools. Conditional external strategy research needs browsing.
A saved example, not a forecast.
The fictional Harbor Pantry run mapped three initiatives to customer outcomes and recomputed the proposed rate thresholds: 90/100 on-time pickups, 90/120 repeat completions and at most 6/200 missing-item complaints.
Original MIT instructions by Pawel Huryn remain unchanged. Only source-guided Codex preparation and bounded artifact checks were completed; no installation, live experiment or business effect was established.
Keep the review honest.
Use authorized material, distinguish facts from targets, agree measurement definitions and validate capacity. A quarter is a planning window, not a delivery promise.