Skip to content

Pilot Guide (4–6 Weeks)

Run a limited, empirical pilot to evaluate whether Dandori improves clarity, flow, and outcomes in your context. Keep the scope small and let data guide next steps.

Why Pilot Before Broader Adoption

  • Reduce risk by validating with one workstream before scaling.
  • Tailor practices to your constraints instead of adopting doctrine.
  • Produce data that informs whether to expand, adapt, or stop.

What to Baseline Before Starting

Capture 4–8 weeks of recent work (or sample 8–12 items) using current ways of working. - Cycle time: Intent → Integration (or request → release). - First‑pass success rate: specs/features passing validation without revision. - Decision latency: time from Decision Request to resolution. - Validation rework rate: % of items that cycle back from Validation to Specification. - Defect escape / operational follow‑up: incidents or fixes discovered post‑release. If data is missing, do a quick sampling review of recent work; approximate consistently.

  • One workstream only (a product area or a single service).
  • 2–4 active specs at a time; WIP stays small.
  • Clear mix: prioritize Clear items; include up to one Complex item.
  • Named roles (can be shared): Spec Owner, Reviewer, Prioritizer.
  • 4 weeks minimum; 6 weeks if your cycle times are longer or work is seasonal.
  • Week 0 is prep: create the Intent queue and visible pipeline; agree on Minimum Viable Dandori.

Suggested Metrics to Observe

Use existing definitions from the framework. - Cycle Time (Intent → Integration) - First‑Pass Success Rate - Decision Latency (tactical ~4h, strategic ≤24h) - Validation Rework Rate - Stage Dwell Time per lifecycle stage (optional) - Pipeline WIP by stage (optional) Track weekly. Prefer trends over single points.

Questions to Ask During the Pilot

  • Clarity: Are intents and acceptance criteria clearer sooner?
  • Decision quality: Are decisions explicit, timely, and visible?
  • Validation readiness: Do specs enable fast, objective validation?
  • Ownership: Are Spec Owner/Reviewer/Prioritizer boundaries working?
  • Coordination: Are handoffs lighter and Integration Reviews producing actions?

What Success Looks Like (Directionally)

  • First‑pass success trending upward or stable with fewer reversals.
  • Decision latency decreasing; fewer “waiting on decision” blocks.
  • Cycle time stabilizing; Stage Dwell Time more balanced across stages.
  • Integration Review surfaces concrete actions; architectural drift decreases.
  • Fewer post‑release surprises; clearer ownership of outcomes. Avoid absolute targets during the pilot; focus on direction and visibility.

What False Positives Look Like

  • Metrics improve while customers feel more pain (e.g., architectural drift increases).
  • Shorter specs that hide missing acceptance criteria.
  • Fewer meetings because nothing ships, not because coordination improved.
  • “Fast” but with growing rework or incident tail.

What Mixed Results Usually Mean (And How to Adjust)

  • Clarity up, latency unchanged: strengthen Decision Requests and decider accountability; review at Integration Review.
  • Faster execution, more rework: improve acceptance criteria and edge cases; add a short Spec Retrospective.
  • Cycle time up, quality up: accept initial investment; keep specs lean and reduce ceremony where data allows.
  • Stage bottleneck moves: adjust WIP or staffing; add/trim ceremony focusing on the bottlenecked stage.

How to Decide: Expand, Adapt, or Stop

  • Expand when: at least one meaningful spec integrates, signals trend in the right direction, and the team sees value.
  • Adapt when: one dimension lags (e.g., decision latency). Change the smallest thing: add Decision Request rigor, tune ceremony frequency, refine templates.
  • Stop when: admin grows without clarity; PM support is absent; decisions remain slow despite explicit requests. Capture lessons and revisit later; don’t force it.

References: Getting Started, Fit and Boundaries, Failure Modes, and Metrics. Use the templates directory for lean artifacts.