How to Plan New Products and Services
A launch plan that is only a list of dates hides the two decisions that matter: when the team stops widening exposure to look at the evidence, and who is accountable for the step nobody has claimed. This playbook runs the work through six phases with the gates written in, and it keeps every gap visible instead of filling it with a plausible guess. Run it here, then install it where the plans actually get made.
Most plans fail quietly rather than loudly. A phase gets skipped because it felt slow, an owner is assumed because only one name was in the room, and a date lands in the schedule because somebody needed a number. None of that looks like a mistake on the day it happens. It looks like a mistake a quarter later, when the rollout widened without anyone deciding to widen it and the metrics that would have caught the problem were never wired up.
The fix is structural. Start from the customer and work backwards to the smallest thing that satisfies the goal, write it down before building, then run the build through phases where alpha to beta and beta to GA are explicit go or no-go rows tied to criteria the team committed to in advance. Where the input is silent the plan says so in the field itself, because an owner nobody named and a date nobody set are the two things a schedule is most likely to invent.
A rollout that widens without a decision row does not have gates. It has momentum.
THE ROUTING, decided once and recorded on line four improvement to an existing flow -> logline brand-new product, big bet -> working-backwards THE SIX PHASES 0 Shaping mechanism, Review 1, design sprint, Review 2 1 Definition job stories, wireframes, MVP plan, Review 3 2 Build bet submission, development, QA, alpha 3 Beta evaluate, go or no-go, instrumentation, beta 4 Launch evaluate, go or no-go, GTM review, GA 5 Learn post mortem, weekly tracking WHAT AN UNFILLED FIELD LOOKS LIKE | 0 | Design sprint | Owner needed | +1 wk | Review 1 |
THE SHAPE
When [situation], I want [motivation], so that [outcome].
SOURCED, and therefore arguable
| 1 | MVP | When I know what I want on a busy morning, I
want to order ahead, so that I skip the line.
Source: support themes, weekly |
UNSOURCED, and therefore labeled
| 5 | Later | Proposed story about group ordering
Source needed |
WHY THE LABEL BEATS A GUESS. A gap is a question somebody can
answer. An invented customer is an answer nobody can check.THE TWO GATES, as rows rather than as intentions
| 3 | Go or no-go, beta | Nadia | Alpha evaluation |
Explicit decision, not a drift |
| 4 | Go or no-go, GA | Nadia | Beta evaluation |
Explicit decision |
ORDER MATTERS MORE THAN THE DATES
instrumentation -> beta rollout evaluable
beta rollout -> instrumentation a feeling
THE DEFAULTS SAY WHAT THEY ARE. Two, two, and six weeks, twenty
in alpha, a hundred in beta. Rules of thumb, labeled in the plan.
Presenting them as this team's measured reality is a fabrication
with a schedule around it.THE FIVE BUCKETS
Metrics Movers | Customer Requests | Customer Delight
Business Needs | Technical Debt
THE CANDIDATE THAT IS NOT READY, kept visible
| Refer-a-friend | Metrics Movers | Not ready to submit |
No artifact and no evidence; cannot be evaluated |
THE ALLOCATION GUIDE, and the departure explained
Bets about 70 percent
Fast experiments about 20 percent
Bugs and debt about 10 percent
A period concentrated in one bucket gets the concentration
surfaced rather than smoothed over. The guide is a guide; a
cycle that departs from it says why on that line.how-to-plan-new-product-and-services/
├── SKILL.md ← the two modes + the delivery boundary
├── meta.yaml ← version, sources, test prompts, changelog
├── LICENSE.md ← MIT
├── assets/
│ ├── milestone-plan-template.md ← the six-phase table + a filled example
│ └── bet-review-template.md ← submission, readout, allocation check
└── references/
├── phase-guide.md ← what each phase is for, and its traps
└── jobs-to-be-done.md ← the job story format, ICE, and sourcing"Use the how-to-plan-new-product-and-services skill to plan this rollout."