Hoshin Kanri X-Matrix

What this tool does

CM8-249 holds a Hoshin Kanri plan as linked entries: breakthrough objectives, the annual objectives that step towards them, the improvement priorities that do the work, and the measures that say whether any of it is happening. Each entry carries an owner, a resource commitment, a status and what it connects to on the adjacent leg. The tool does the book-keeping — direction-aware attainment, entries that connect to nothing, priorities nobody funded — and prints a review pack.

Everything runs inside this single file — no account, no upload and no network request of any kind, so your strategy, your targets and your shortfalls never leave this computer.

What Hoshin Kanri is

Hoshin Kanri is strategy deployment: the discipline of choosing very few objectives and connecting every piece of improvement work to one of them. The name and the method come from Japanese manufacturing in the twentieth century, and are often rendered as "policy deployment".

The problem it solves is unglamorous and universal. Most organisations have no shortage of strategy; they have a shortage of connection between the strategy and what people actually did last week. Departments improve what is convenient, and at the year end a great deal of effort has moved nothing the leadership said mattered. Hoshin Kanri forces four things onto one page: what must change over several years, what changes this year, what work delivers it, and what number proves it. Anything outside that chain is not part of the plan — which is the point, and the part people find hard.

The X-matrix, explained plainly

The classic artefact is the X-matrix: a large grid with the four legs written along its edges — breakthroughs at the bottom, annual objectives at the left, priorities along the top, measures at the right — and correlation marks in the corner blocks where two adjacent legs meet. A mark in a corner says "this priority serves that annual objective", so the whole strategy and its links are visible at once.

This tool does not draw that grid, and it is worth being straightforward about why. A real X-matrix is a wall-sized document; at the width of a laptop, let alone a phone, it is unreadable, and the corner marks — the actual content — disappear first. So the same information is held here as linked entries: one row per element of one leg, with a Links to field naming what it connects to on the adjacent leg. The corner correlations are those links, in words rather than dots — which buys something paper cannot, because the links become checkable.

The vital few

If everything is a priority, nothing is. The standard advice — largely intact after decades of practice — is three to five breakthrough objectives and no more, with similar restraint on the annual objectives beneath them. The settings record the limit you set yourself, and the tile says when the plan has drifted past it.

The same restraint applies to the priority leg, and that is where plans usually fail. Twelve improvement priorities is not ambition; it is a to-do list with a cover sheet. Twelve cannot all be resourced, so they are all half-resourced and all run slowly, and at the year end you have twelve things at sixty per cent and nothing finished. Four properly staffed will beat twelve. Deciding what not to do is the work, and the first honest pass is uncomfortable — every candidate has a sponsor.

Catchball

Catchball is the negotiation that turns a plan into commitments, and it is the part most often skipped. The name is literal: leadership throws a draft objective down to the level that would have to deliver it; that level throws back what it would take and what it cannot do. The ball goes back and forth — usually two or three rounds — until both sides agree an objective, a target and the resources behind it. Nobody ends up committed to something they told you was impossible.

It is not consultation theatre, and not consensus-seeking. Leadership still decides the direction; catchball decides whether it can be delivered by the people who would have to deliver it. The most valuable output is often a target that came back lower than the one thrown down, with a credible reason — a target people believe in beats a higher one they have privately written off.

The Agreed in catchball status exists for this. It marks the moment an entry stopped being leadership's draft and became somebody's commitment. A plan sitting entirely in Draft has been negotiated with nobody; one whose breakthrough leg is Active while the priority leg is still Draft has been announced rather than deployed.

Linking discipline

Every annual objective must serve a breakthrough objective. Every priority must serve an annual objective. Every measure must measure one. Those three rules are the integrity of the method, and the Unlinked entries tile is the check on them.

An entry that connects to nothing is the core failure mode of strategy deployment, and it is almost never malicious. It is the good idea somebody brought to the planning session, the project already running, the audit programme everyone agrees is sensible. It gets in because refusing it feels petty — then it consumes the same people and weeks as the work that does serve the strategy.

When the tile is above zero there are two honest choices: name the annual objective it serves, which clears the flag, or accept that it is business-as-usual work and manage it outside the Hoshin plan. What is not fine is leaving it in, unlinked, competing for resources with the vital few.

Resourcing honestly

The most common way a strategy quietly fails is not disagreement, and not bad objectives. It is a priority everyone agreed to and nobody funded. It stays green for a quarter because nothing has gone wrong yet, then amber because it has not started, then is found in month nine exactly where it was in month one.

Resourcing means people, money and — most often the missing one — protected time. "The team will fit it around the day job" is not a commitment; it is a hope, and it loses every week to whatever is shouting loudest. So the tool refuses a priority marked Active with a resource commitment of None yet. That combination is a wish, and writing it down as a plan is how it survives to the year end unchallenged. Either commit the resource or move the status back to Agreed, which honestly says: negotiated, not started. If the donut is mostly red, you have too many priorities.

Measures that are outcomes, not activities

A measure belongs on the measure leg only if it tells you whether the objective is being achieved. "Training sessions held" is an activity: it rises while nothing changes. "Stations with three signed-off operators" is an outcome. The test: if it could hit its target while the objective it links to made no progress, it is the wrong measure. Every measure needs a target, which is why the tool will not save one without it.

Attainment uses the direction you set, so 100 always means "exactly on target" and more is always better, whether the measure is meant to rise or fall:

A quote turnaround of 4.5 days against a 2-day target scores 2 ÷ 4.5 × 100 = 44.4 — the shortfall reads the same way as for a measure pointing the other way. Where a value is missing or either is zero the division is meaningless, and the tool shows — rather than a confident wrong number; a "lower is better" target of zero will not save at all.

The monthly review

Hoshin Kanri is a yearly plan reviewed monthly. The review is short and has one job: find the priorities drifting while there is still time to act. Update the current values, set the status honestly, and spend the meeting on the Integrity and risk table rather than reading the matrix out.

When a priority is at risk there are four honest outcomes, and "keep an eye on it" is not one of them. Give it more resource; cut its scope so the reduced version still serves the annual objective; change the annual objective because the world moved; or drop it and say so. Dropping a priority in month four and redirecting its people is a decision. Carrying it at amber until December is not.

How this relates to OKRs and the balanced scorecard

OKRs share the objective-plus-measure structure and the cascade, but run on a quarterly rhythm with goals usually set bottom-up and deliberately stretched, where Hoshin runs annually against a multi-year breakthrough and treats the target as a commitment rather than an aspiration.

The balanced scorecard is a measurement framework: it insists your measures cover financial, customer, process and capability perspectives, but says nothing about which improvement work will move them — the gap the priority leg fills.

FAQ

How many entries should a plan have? Typically three to five breakthroughs, three to five annual objectives, four to six priorities and one or two measures per annual objective — fifteen to twenty in total. Much larger and the plan is a work list.

More than one plan year in one file? You can, and the search box isolates a year, but one file per plan year keeps last year's dropped priorities out of this year's counts.

What if a priority serves two annual objectives? Name both in the Links to field, comma-separated — that is what the corner blocks express. Be suspicious of one claiming to serve four.

Do departments each need their own matrix? Beyond about a hundred people, yes: the level below takes your annual objectives as its breakthroughs and runs its own catchball.

Saving your work

Entries, settings and the report header are written to this browser's local storage as you type. That storage belongs to one browser on one computer: another browser, a private window or a tool that clears site data will not have it.

Treat Export .json as the real save — one file containing everything, which Import .json restores anywhere. Export CSV gives you the entries for spreadsheet work. Reset asks twice, then erases everything, with no undo. A Hoshin plan states your strategy and your shortfalls on one page — treat exports as confidential.

Accuracy & disclaimer

The arithmetic here is simple and the tool does it faithfully. What it cannot do is judge the plan: whether the breakthrough objectives are right, whether a link is real or merely asserted to clear a flag, or whether a target is ambitious or soft. A fully linked, fully resourced matrix can still describe a bad strategy tidily.

The matrix records the plan and its links. The catchball conversations that turn it into commitments, and the monthly reviews that catch a priority while it can still be saved, happen outside this file — all the tool can do is make their absence visible. This is a planning and review aid, not strategic or financial advice.