roadmap-nownextlater

name	roadmap-nownextlater
description	Use when the backlog needs to become a defensible plan — to sort bets into Now / Next / Later by evidence and capacity, keep Now small, and re-rank every cycle against what the loop returns.

Now / Next / Later Roadmap

A Now / Next / Later roadmap communicates priority and confidence without pretending to know exact dates you can't hit. Now is committed and small; Next is likely and shaped; Later is directional and cheap to reorder. The discipline is keeping Now honest — a roadmap where everything is "Now" is just a wish list with a deadline.

Steps

1. Collect every bet in one shape. For each candidate: the outcome it targets, the evidence it's worth doing (a loop signal, a metric, a customer pattern), and a rough size (small / medium / large). A bet with no evidence line goes back for a signal, not into the plan.
2. Score by evidence and impact, discount by uncertainty. Rank bets by how much they'd move a real number, then discount ones resting on opinion rather than signal. Loud internal preference is not evidence.
3. Fill Now against real capacity. Put only what the team can actually finish this cycle into Now — leave slack for the reliability and support work that always arrives. Overstuffing Now is how everything slips together.
4. Shape Next, sketch Later. Next holds the small set you're fairly confident comes after Now, shaped enough to start soon. Later is a directional list, deliberately loose — its job is to say "we see these and they're not forgotten," not to commit.
5. Name what you're not doing. Explicitly list bets you considered and deprioritized, with a one-line reason. Un-decided requests fester; a recorded "not now, because" ends the lobbying.
6. Re-rank every cycle. Move items across horizons based on what the loop actually returned — activation, retention, pipeline — not on how long something has waited. Age is not priority.

Quality bar

• Now fits real capacity with slack; it is small enough to actually finish.
• Every bet carries an evidence line; opinion-only bets are discounted or parked.
• Deprioritized bets are named with reasons, not silently dropped.
• The roadmap is re-ranked each cycle against measured outcomes.

Output format

NOW (committed, fits capacity):
  - <bet> — outcome <x> — evidence <signal> — size <s/m/l>
NEXT (shaped, likely):
  - <bet> — outcome <x> — evidence <signal>
LATER (directional, reorderable):
  - <bet> — outcome <x>
NOT DOING (this cycle):
  - <bet> — because <one line>
Re-rank note: <what moved and why, from which loop signal>