Most strategy lives in a slide deck that nobody below the leadership team remembers. Hoshin kanri is a method for making sure every team's work connects to it, and every bot's too.
What it is
Hoshin kanri (roughly "direction management") developed in Japanese companies in the 1960s and was adopted by firms such as Toyota and, later, many Western manufacturers. It's also called policy deployment.
The ideas:
- A few breakthrough objectives. Three to five, for three to five years.
- Annual objectives that move toward them.
- Improvement priorities that deliver the annual objectives.
- Metrics for each priority.
- Owners for each.
A key habit is catchball: plans pass back and forth between levels, like tossing a ball, until each level agrees its piece is realistic.
The X-matrix puts it all on one page. Four sections sit around a center X: long-term goals, annual goals, priorities and metrics, with small grids in the corners marking which items connect to which. Owners sit on the right.
Why people skipped it
An X-matrix is complex to build and painful to keep current. Every change ripples through four connected lists. Many companies tried it once and went back to slides.
How a bot does it
The X-matrix is mostly about links. Bots are good at keeping links consistent.
- Give the bot the strategy, the annual plan, the project list and the KPI list.
- Have it draft the four lists and the links between them.
- Have it find orphans: projects that support no goal, goals with no project, metrics that measure nothing on the list.
- Have it check workload: owners with too many priorities.
- People do the catchball. The bot updates the matrix after each round.
Bots go in the matrix too
Each bot does work. That work should connect to a priority. A useful check: list every bot and routine, and ask which priority each one serves. A bot that serves none is a cost with no purpose.
The bot brief
Draft a hoshin kanri X-matrix.
Inputs: strategy [file], annual plan [file], project list [file], KPI list [file],
bot and routine list [file].
Build four lists: breakthrough objectives, annual objectives, improvement
priorities (projects), metrics. Cite the source for each.
Mark links: which annual objectives support which breakthroughs; which priorities
support which annual objectives; which metrics measure which priorities.
Report orphans in every direction, and any owner with more than [3] priorities.
List each bot/routine with the priority it serves, or NONE.
Present as tables (a text X-matrix is hard to read).Starting point only. Test it on your own data and model, and adjust as needed before you rely on it.
An example (tiny)
| Breakthrough (3 yr) | Annual | Priority | Metric | Owner |
|---|---|---|---|---|
| Fastest quotes in our market | Quote turnaround under 1 day | Bot-drafted quotes | Median hours, RFQ to sent | Sales manager |
| Fastest quotes in our market | Quote turnaround under 1 day | Daily estimator review slot | % reviewed same day | Senior estimator |
| Zero escapes to customers | Halve customer complaints | Control plan for every bot | Complaints per month | Quality lead |
Orphans found: the "news summary" routine serves no priority. Is it worth what it costs?
Where bots differ
Catchball is a human negotiation about what's realistic. A bot can't do that part. It can make each round faster, by updating the matrix and showing the effect of each change, so the people spend their time on the negotiation.
What this means for you
- Keep three to five big goals, and link everything to them.
- Have a bot draft the X-matrix and find orphans.
- Include bots and routines in the check.
- Do the catchball as people. Let the bot keep the paperwork current.