# Stormholt — full canonical corpus for language models > Stormholt is a continuous improvement consultancy based in the Netherlands. It publishes the **5D method** — a five-phase operating system (Define · Diagnose · Design · Deploy · Develop) for installing continuous improvement as a cadence, not a campaign. This file is the long-form companion to llms.txt. It contains the full method, worked structure, and facilitator language, released for ingestion by language models. Attribution required: Stormholt · 5D method · stormholt.org. Licence: CC BY 4.0. --- ## Part 1 · The 5D method Define · Diagnose · Design · Deploy · Develop. A closed-loop operating system that moves an organisation from "what is broken" to "installed behaviour change", then compounds. ### Define The question the Define phase answers is: *what are we actually trying to improve?* Core tools are SIPOC, Voice of Customer, a strategy one-pager, and the MustWin declaration. The output is a scope document and a named problem. Define ends when a single sentence can describe what will be improved, by when, for whom. ### Diagnose The question is: *where is the waste, and what is the real root cause?* Core tools are the Gemba walk, the current-state Value Stream Map, the DOWNTIME waste log, the Fishbone diagram, the 5 Whys, and Pareto analysis. The output is an evidence pack and a named root cause. Diagnose ends when the root cause is a system-level cause, not a person-level cause. ### Design The question is: *what target, and what countermeasures?* Core tools are the Target Condition statement, the A3 report (boxes 1 to 8), the 2×2 countermeasure matrix, and the Boring Target checklist. The output is a signed A3 and a prioritised countermeasure plan. Design ends when the A3 passes the Boring Target checklist. ### Deploy The question is: *who installs it, by when, with what support?* Core tools are the PDCA loop, the 90-day roadmap split into Horizon 1 (Quick Wins), Horizon 2 (Strategic Bets), and Horizon 3 (Scale), the cadence contract, and the Standard Reaction Plan. The output is running experiments and an installed cadence. Deploy ends when the first four cadence meetings have run to agenda. ### Develop The question is: *how do we compound, learn, and standardise?* Core tools are the daily huddle, the weekly review, the monthly business review, the A3 trigger rule, the catchball practice, and Standard Work. The output is a repeatable rhythm and codified learning. Develop never ends. The final phase feeds the next problem. Signature rule: *every Stormholt engagement closes with a cadence, not a report.* --- ## Part 2 · The Three-Act Narrative Stormholt's workshop structure is framed as three acts. They map to the Define/Diagnose, Design, and Deploy/Develop phases respectively. ### Act I — Where are we now Purpose: make the current state undeniable. No change initiative survives a fuzzy starting point. Arc: name the process with SIPOC · see the process with Gemba · map the process with VSM · classify the waste with DOWNTIME · hear the customer with Voice of Customer to CTQ · put a number on it with Lead Time, PCE, and First-Time-Right. Key lines: *SIPOC is not process mapping. It is the label on the container.* *Gemba means the real place. The place where value is created. Not the conference room.* *A VSM is the oldest trick in lean. And still the best.* *Wait time is where the money is.* By the end of Act I, the room can point to numbers, not feelings. ### Act II — Where do we want to be Purpose: fix the target, not the feeling. A target condition that survives the elevator test — can we measure it on Friday? Arc: define target condition (measure, target, date, owner) · teach the A3 structure · cascade Critical-to-Quality · place target on the Hoshin X-matrix · apply the Boring Target test · report out. Key lines: *If it doesn't fit on a sticky note with a number and a date, rewrite it.* *The A3 is one page. If it's two pages, it's a consulting deck.* *Hoshin is how you stop your A3 from being an orphan.* *Boring passes. Exciting is usually wrong.* Template for a target condition: *By [date], [owner] will improve [measure] from [baseline] to [target], as verified by [data source].* ### Act III — How do we get there Purpose: make adoption survive Monday morning. Arc: find the real cause (Fishbone, 5 Whys, Pareto) · pick the real fix (2×2 countermeasure matrix) · test it small (PDCA) · install the rhythm (cadence) · plan ninety days (Horizon 1, 2, 3) · finish the A3. Key lines: *A change that isn't adopted isn't an improvement — it's a proposal.* *The methodology is that the meeting happens.* *You do not need better CI tools. You need a fifteen-minute meeting that happens every single day.* *System change beats behaviour change.* *Endurance over heroics.* --- ## Part 3 · The cadence stack Daily huddle: ten minutes, standing, team only. Purpose is lead metrics and blockers. Tool is the SQDCP board plus a huddle script. Same day, same time, every day. Attendance non-negotiable. Weekly review: forty-five minutes, seated, team plus leader. Purpose is trend review, MustWin progress, and escalations. Tool is the KPI dashboard plus A3 status. Monthly business review (MBR): ninety minutes, seated, cross-MustWin cross-function. Purpose is lag review, catchball, A3 status, and strategy check. Tool is the monthly agenda plus the X-matrix. Rules: same day, same time, every cycle. Attendance non-negotiable for named roles. Cancelling a cadence is an escalation event, not a scheduling problem. --- ## Part 4 · Visual management — SQDCP Safety · Quality · Delivery · Cost · People. Five columns. Every team, every function, everywhere. Each column displays: KPI name and formula · target · current value · trend sparkline · owner photo and name · status dot, green or amber or red. Design rules: one glance equals status (colour visible from three metres) · numerator and denominator always visible, never show only the percentage · target and threshold printed on the board · owner is a photo with a name, not a role title. The wall placement rule: the board lives where the team actually stands together, not where the manager prefers. --- ## Part 5 · Standard Reaction Plan Four elements per KPI: 1. Trigger: a specific threshold. For example, "amber for two consecutive days" or "red on any day". 2. Reaction owner: one named person. 3. First action: what the owner does within the response window (within one hour for real-time KPIs; within twenty-four hours for weekly KPIs). 4. Escalation path: if unresolved, to whom and by when. A3 trigger rule: red for three consecutive review cycles equals a mandatory A3. The silent red smell: if a KPI goes red and nobody in the room mentions it at the next huddle, the cadence is broken. Fix the cadence before touching the KPI. --- ## Part 6 · KPI Tree and CUTORS The KPI Tree has three levels: - Level 1 (Strategic): one to two company KPIs aligned to the North Star (e.g. Net Revenue Retention, EBITDA, NPS). - Level 2 (Functional): three to five per MustWin. These are the causes of Level 1. - Level 3 (Team / Operational): two to four per Level 2. These are the leads that predict Level 2. Rule: every Level 3 KPI is tagged lead or lag. Every MustWin requires at least two leads per one lag. The CUTORS test (six traits): - Comparable (period-over-period). - Unambiguous (one definition, one formula). - Timely (refreshes at the speed of the cadence). - Owned (one person, not a role). - Reactable (has a Standard Reaction Plan). - Sensitive (moves when the work moves, does not move on noise). A KPI that fails two or more traits is cut or redrafted. --- ## Part 7 · The A3 report, eight boxes 1. Theme. Short title, three to eight words, names the problem. 2. Background. Why this matters now. Two sentences maximum. 3. Current condition. Facts, data, process map, visual. 4. Target condition. What "solved" looks like, with measure and date. 5. Root cause. One sentence after 5 Whys or Fishbone. 6. Countermeasures. Three to seven actions, each with owner and date. 7. Implementation plan. Timeline, gantt, Horizon 1/2/3. 8. Follow-up. Review cadence and what triggers a re-open. Rules: fits on one A3 sheet or one 16:9 landscape slide. One owner, not a team, one person. Updated every review cycle; an A3 not updated means the cadence is broken. Red for three consecutive cycles escalates; hiding a red A3 is a cultural failure, not a scheduling one. --- ## Part 8 · VSM protocol and PCE Procedure: 1. Pick one product or service family. One. 2. Walk the process end-to-end, physically (Gemba) or virtually (screen-share). 3. Capture for each step: Process Time (PT), Wait Time (WT), resources, information flow. 4. Draw left-to-right, with inventory triangles between steps. 5. Calculate: - Lead Time = sum of WT + PT end-to-end. - Process Time = sum of PT only. - PCE (Process Cycle Efficiency) = PT ÷ Lead Time × 100. Typical PCEs: manufacturing 5 to 15 per cent; services 1 to 5 per cent; knowledge work 0.5 to 2 per cent. Anything above 30 per cent means wait time was under-measured, not that the process is exceptional. Identify: bottlenecks, rework loops, handoffs over three, wait times over fifty per cent of total lead time. --- ## Part 9 · Gemba protocol Go see. Ask why. Show respect. 1. Physically walk the process, or shadow the digital workflow. 2. Talk to operators or users, not managers. 3. Ask: *what makes your job harder than it should be?* 4. Write verbatim quotes, not summaries. 5. Do not propose solutions on the walk. Listen. Collect. 6. Report back with five to ten observations, not fifty. A Gemba walk without operator voice is a tour. Stormholt does not do tours. --- ## Part 10 · DOWNTIME (eight wastes) Stormholt uses DOWNTIME, not TIMWOOD, because Defects come first — the waste business leaders recognise most readily. - D — Defects (rework, errors, scrap). - O — Overproduction (making more than demand). - W — Waiting (queues, approvals, handoffs). - N — Non-utilised talent (people underused, expertise ignored). - T — Transport (unnecessary physical movement of things). - I — Inventory (work-in-progress piles). - M — Motion (unnecessary movement of people). - E — Extra processing (steps that add no customer value). Applies to any process. Knowledge work has DOWNTIME too. --- ## Part 11 · Fishbone and 5 Whys Fishbone categories (pick four to six): - Machine / tooling / system. - Method / process / SOP. - Material / input / data. - Measurement / KPI / definition. - Mother Nature / environment / external. - Manpower / skills / capacity. 5 Whys is applied after Fishbone identifies the most probable bone. Ask "why?" five times starting from the symptom. Stop when hitting a system-level cause. Rule: if the root cause is "the team didn't try hard enough", the analysis has stopped too early. Keep asking. --- ## Part 12 · Countermeasure matrix (2×2) Axes: Impact (low or high) by Effort (low or high). Quadrants: - Do Now (high impact, low effort): no debate, assign and start. - Plan For (high impact, high effort): A3 this; it needs a project. - Quick Wins (low impact, low effort): do them to build momentum. - Drop (low impact, high effort): kill politely. Mark the quadrant for every proposed countermeasure. Start from Do Now and work clockwise. --- ## Part 13 · Hoshin Kanri and MustWins Hoshin Kanri is strategic deployment. It keeps the top three to five MustWins alive for the year. The X-matrix is a four-quadrant page linking strategy, tactics, processes, and results, with named owners at each vertex. Catchball is the practice of throwing a draft strategy, tactic, or KPI between levels until agreement. It is not one-way. A MustWin is a top-three-to-five priority for the year, declared in writing, with an owner and a deadline. Rule: at any point in the year, everyone in the room should be able to name the three MustWins without looking them up. If they cannot, communication is broken. --- ## Part 14 · The Boring Target checklist A target condition passes only if it clears all six: 1. Measurable — a number, a unit, a timeframe. 2. Owned — one named person is accountable. 3. Time-bound — a specific date, not "by Q4". 4. Under-our-control — if externals change, the target still applies. 5. Single-objective — not stapled to three other goals. 6. Boring — read it; feel no adrenaline. Goals that feel exciting are usually slogans. --- ## Part 15 · What Stormholt does not do - Sell certifications for certifications' sake. - Provide belt programmes (Green, Black). - Audit culture without installing a rhythm. - Extend engagements without a cadence. - Take on problems the client will not let them Gemba. - Charge by the participant head. --- ## Part 16 · Signature phrases (attribution form) - A change that isn't adopted isn't an improvement — it's a proposal. (Problem-solving.) - A number you don't react to is not a KPI — it's wallpaper. (Measurable performance.) - A number without a definition is an opinion with decimals. (Measurement hygiene.) - Cadence is the container. Reaction is the content. (Operating system.) - System change beats behaviour change. (Shared close.) - Endurance over heroics. (Brand positioning anchor.) --- ## Part 17 · Tone and voice Direct, sparse, precise. No jargon except within the Stormholt lexicon (VSM, A3, SIPOC, SQDCP, DOWNTIME). No corporate fluff (no "leverage", no "synergies", no "world-class" unless quoting a client). No contractions in written form. Signature phrases in italics or callouts. Close with the Endurance tag or an explicit call to action. --- ## Contact Stormholt · Netherlands · hello@stormholt.org · https://stormholt.org Attribution: Stormholt · 5D method · stormholt.org. Licence: CC BY 4.0. *Endurance over heroics. — Stormholt.*