Hoshin planning is usually described as a set of artifacts: an X-matrix, catchball, a bowler chart. It is better understood as a sequence of seven stages, each with an output the next stage depends on. Skip one and the error shows up months later.
In this guide
- Why sequence matters
- Stage 1: Reflect
- Stage 2: Re-examine the breakthroughs
- Stage 3: Draft annual objectives
- Stage 4: Select the vital few
- Stage 5: Correlate and gate
- Stage 6: Catchball
- Stage 7: Lock and deploy
- A worked example: one objective through all seven stages
- What the cycle needs from the software
- Supporting the cycle in software
- Questions this article answers
Key Takeaways
- →Seven stages, each with one output. The order is the method, not a formality.
- →Stage 1 is the one most often skipped, and the one the rest depends on. Without verdicts on last year, losing bets come back with new names.
- →Draft annual objectives from the breakthrough trajectory, not from last year plus an increment.
- →Stage 4 is subtraction. A priority list with no stop-doing list beside it is a wish list.
- →The plan is finished when the first monthly review can run from it. Not when it looks complete.
- →Most strategic planning software supports one stage well. Ask which of the seven it holds before a shortlist forms.
Why sequence matters
Most planning seasons have the right ingredients in the wrong order. Targets get set before breakthroughs are re-examined. Correlations are drawn before priorities are chosen. Catchball starts before the matrix can be defended, so the negotiation is about wording instead of commitments.
The result is a matrix that looks complete in February and starts coming apart in March. Each problem traces back to a stage that was rushed, merged with another, or skipped. The cycle below makes each stage explicit and gives it a single output. You can tell at any point whether the plan is ready to move on.
Stage 1: Reflect
The cycle begins with the year that is ending. For every primary correlation on the current X-matrix, the planning team records a verdict: held, partly held, lost, or too early to tell. For targets that were renegotiated during the year, it records why. For priorities that consumed far more effort than planned, it records that too.
Output: a reflection summary, a few pages at most, that the planning offsite opens with.
How it fails: it’s cut for time. Or reduced to a slide of red, amber and green with no verdicts. The detail is in Hansei: reviewing last year’s X-matrix.
Stage 2: Re-examine the breakthroughs
Breakthrough objectives span three to five years. Each cycle, the executive team checks each one against its trajectory: is it where it needs to be by now, and is it still the right ambition? Most breakthroughs are kept. Some are adjusted. Occasionally one is retired, deliberately and on the record.
Output: confirmed breakthroughs, each with where it must be by the end of next year.
How it fails: breakthroughs are carried forward without question, or quietly rewritten so last year’s shortfall disappears. Breakthrough and annual objectives in one system covers the link between the two horizons.
Stage 3: Draft annual objectives
Annual objectives are drafted from the breakthrough trajectory: what must be true by December for each breakthrough to stay on course. This is different from taking this year’s results and adding a reasonable increment. It usually produces more demanding numbers.
Output: draft annual objectives with proposed targets.
How it fails: targets anchor on this year’s results and drift towards incrementalism. See Setting targets from the breakthrough trajectory.
Stage 4: Select the vital few
Priorities are the work that will move the annual objectives. The discipline here is subtraction: choosing few enough that each gets real attention, and deciding what stops to make room. A capacity check against the people who will do the work closes the stage.
Output: a short priority list, a stop-doing list and a capacity check.
How it fails: every proposal makes the list, and nothing stops. See Choosing the vital few and Stop-doing lists.
Stage 5: Correlate and gate
Now the X-matrix takes shape. Each priority is correlated with the KPIs it is meant to move, each correlation is graded, and every primary correlation states its mechanism, expected contribution, lag window and kill signal before it’s accepted.
Output: a draft X-matrix in which every dot can be defended.
How it fails: dots are drawn so no priority looks orphaned. The gate is described in The X-matrix is a set of hypotheses.
Stage 6: Catchball
The draft goes to the level below for negotiation. Each objective’s owner proposes; the receiving level responds with what it can commit to and what it needs; the proposal is revised. Two or three rounds are normal. The rounds should run in parallel across objectives, with the state of each one visible.
Output: agreed targets, each with its reasoning and any revisit trigger recorded.
How it fails: catchball runs by email, sequentially, and eats the first quarter. See Running catchball rounds.
Stage 7: Lock and deploy
The final stage turns the agreed plan into an operating plan. Monthly plan values are phased to match how each result is expected to move. Every row has one owner. The check-in cadence and review calendar are set. The matrix is locked, which means changes now go through the quarterly review rather than ad hoc edits.
Output: a locked matrix the January review can run from.
How it fails: the plan is “final” but not operable: annual targets only, shared owners, no cadence. See Locking the plan.
A planning cycle is finished when the first monthly review can run from its output. Not before.
A worked example: one objective through all seven stages
Here is a single illustrative objective, services as a share of sales, followed through the cycle.
| Stage | What happened to the services objective |
|---|---|
| 1. Reflect | Verdict on last year’s service-network bet: partly held. Centers opened, but adoption lagged until call routing was changed in Q3 |
| 2. Breakthroughs | Breakthrough kept: 40% of sales by end 2028. Currently 31% |
| 3. Annual objectives | Required position for end 2027: 34%, working back from the trajectory |
| 4. Vital few | Two priorities chosen: complete the service network and launch multi-year contracts. Old dealer-service scheme stopped |
| 5. Correlate and gate | Both priorities graded primary, each with a mechanism, contribution, lag window and kill signal; contributions fit the three-point gap |
| 6. Catchball | North America committed to 31%, rising to 36% if the contract model is live by 31 March |
| 7. Lock and deploy | Ramp phasing agreed; contract attach rate added as the leading KPI; first check-in due 31 January |
Illustrative.
Notice how much of the final plan came from stage 1. The reflection’s lesson, that the network only worked once routing changed, became an adoption measure in stage 7. Without it, this year’s plan would have measured the second wave of centers by how many opened, and repeated last year’s lag.
What the cycle needs from the software
Each stage leaves something behind that the next one reads. That is the test to apply during an evaluation, because most strategic planning software supports one or two stages well and treats the rest as a document you attach. The gap shows up a year later, at the next cycle, when stage 1 has nothing to read.
| Stage | What it needs the tool to hold | How to test it in a demo |
|---|---|---|
| 1. Reflect | Last year’s correlations with a verdict on each, and why targets changed | Ask to see a matrix from a closed year, with the reasons |
| 2. Breakthroughs | Objectives on a multi-year period, with cumulative movement | Ask to see a 2026 to 2028 objective and how far it has moved |
| 3. Annual objectives | The trajectory visible while the target is typed | Ask what is on screen when next year’s number is entered |
| 4. Vital few | A stop-doing list that is part of the plan, not a separate slide | Ask where a retired priority goes and whether it stays visible |
| 5. Correlate and gate | Correlations as records with grade, mechanism, lag and kill signal | Click a dot and see what opens |
| 6. Catchball | Parallel threads per objective, with the state of each visible | Ask to see six negotiations running at once |
| 7. Lock and deploy | Monthly phased plan values, one owner per row, a cadence | Ask whether a plan row can be phased, or only an annual target set |
Seven stages, seven questions. A tool that answers two of them is a drafting surface.
Stage 7 is the one that separates a planning tool from a strategy execution platform. A product that takes an annual target but will not phase it by month cannot produce an operating plan, so the January review falls back to slides and the whole cycle ends one step short. Ask that question early; it is cheap to answer and expensive to discover.
Stage 1 is the one buyers forget, because it only matters at the next cycle. If the tool keeps no history, your first planning season looks fine and your second starts from a blank page again. The seven pass/fail tests cover the execution side, and the cost article covers what to total before comparing quotes.
Supporting the cycle in software
Each stage depends on something software can hold: last year’s record for stage 1, multi-year objectives for stage 2, graded correlations for stage 5, catchball sessions for stage 6, phased plan values for stage 7. In Profit.co these live in one place. The X-matrix can be viewed across years, objectives can carry multi-year strategy periods, correlations are graded records, catchball opens on each objective, and the bowler chart takes monthly plan values.
Hoshin Kanri sits in the Essentials plan with OKRs and Goals, Balanced Scorecard and Strategy Roadmap, so the team OKRs that carry a priority into the quarter run on the same data model as the breakthrough above them. One subscription covers the whole cycle. The alternative is a planning tool, then OKR software for the quarter, then balanced scorecard software when the board asks, with stage 1 spread across three systems that keep separate histories. The 30-day free trial is long enough to load last year’s matrix and run stage 1 on it, which is the stage worth testing first. More on the hoshin kanri software overview.
Free Product Demo
Plan next year from this year’s record
A 30-minute walkthrough, tailored to your team.
Book a free demo
What happens in the call
How your teams set goals, track progress and review results.
Strategy, performance, projects and engagement on one platform, with AI.
A practical path to align teams, execute faster and drive results.
Questions this article answers
Reflect on last year’s plan, re-examine the breakthrough objectives, draft annual objectives, select a short list of priorities, correlate priorities with KPIs and gate them, negotiate targets through catchball, and lock and deploy the plan.
When the first monthly review of the new year can run from it: every row has one owner, KPIs have baselines, plan values are phased, and the check-in cadence is set.
The reflection on last year’s plan. Without it, the planning room can’t tell which correlations held, so losing bets tend to return with new names.
Two stages are unusual: a gate that makes every correlation state its mechanism and kill signal before it is accepted, and catchball, where the level below negotiates the target rather than receiving it.
All seven stages, not two. History for the reflection, multi-year objectives, the trajectory visible at target setting, a stop-doing list in the plan, correlations as records, parallel catchball, and monthly phased plan values.
Long enough that catchball gets two or three rounds and short enough to finish before the year starts. The constraint is stage 6, which runs in parallel across objectives rather than one at a time.
