Cadence Is the Easy Half of Leader Standard Work
Most organizations that decide to install leader standard work start in the same place: the calendar. A daily walk, a weekly review, a monthly meeting, a quarterly evaluation. Intervals get chosen, invitations go out, and the routine exists.
A year later the routine is still running and the organization cannot answer a question with it.
Not because the cadence was wrong. Because nothing came out of it that could be used twice.
Cadence is the part everyone gets right
Choosing an interval is a low-difficulty design decision. Weekly, monthly, annually — the answer follows from volume and from how fast the thing being watched can change. It takes an afternoon and it is rarely the reason an install fails.
The reason it looks like the hard part is that it is the visible part. A cadence is something you can point at. It appears on a calendar, it produces attendance, and it feels like the system is running.
What comes out of the event is where the design work actually is, and it is the half that usually gets left to whoever is running the routine that week.
An output is an entry, not a record
This is the reframe the rest of the model depends on.
The instinctive view is that a leader standard work event produces a record — evidence that the review happened, dated and filed. Under that view a notebook page is sufficient. The event occurred, something was written down, the obligation is met.
The useful view is that the event produces an entry in a database of past performance.
The distinction is not academic. A record answers whether something happened. An entry participates in something larger than itself. One entry is nearly worthless on its own. A hundred entries in the same shape is a trend line, and a trend line is the only thing that supports a decision about direction rather than a decision about today.
Which means the output has a job beyond documenting the event, and that job sets the requirements. It has to be comparable to the last one. It has to be usable by the activities around it. And it has to be able to travel upward. A quality management system that produces evidence of activity but no accumulated picture of performance is doing the expensive half of the work and skipping the half that pays.
Vertical consistency: same activity, same shape, every time
The first requirement is the one people nod at and then do not implement.
If the same activity produces a differently structured output each time it runs, nothing can be aggregated. The information may all be present. It is present in a form that requires a human to read every instance and reconstruct the pattern by hand, which means in practice nobody does it, and the pattern goes unnoticed until it becomes an event.
The test is simple. Take twelve months of outputs from one activity and try to answer a question that spans them. Is this getting better or worse. Where is the recurrence. What changed after the last intervention. If answering requires opening all twelve and reading them, the output was shaped as a record.
This is the same failure that makes an internal audit schedule produce a stack of reports rather than a picture of the system. The schedule is defined, the audits happen on time, and the findings never accumulate into anything because each report was built to stand alone.
Lateral coherence: outputs that have to reach each other
The second requirement gets missed more often, and it is the one that separates a set of routines from a system.
Different leader standard work activities are not independent. Consider the nonconformity and corrective action process — a review of the register for trends, run at whatever frequency the volume justifies. Those trends are precisely the input that tells the organization whether a risk is new, whether an existing risk has changed in likelihood, or whether a control that was assumed effective is not performing.
So the output of the corrective action review is an input to the risk activity. That is a handoff, and handoffs have a format requirement.
If the corrective action output is shaped in a way the risk activity cannot ingest — narrative where the risk review needs categories, or categories that do not map to how risks are described — the handoff does not happen. The two activities continue running on schedule, both producing outputs, neither informing the other. The risk review then reconstructs the picture from what people happen to remember, which is a smaller and more flattering picture than the register would have given it.
Outputs that cannot reach each other do not fail loudly. They pool. Each routine looks healthy in isolation and the connective tissue never existed.
Vertical rollup: management review is the top-altitude instance
The same logic runs upward, and this is where the model becomes a management system rather than a set of manager habits.
Risk evaluation, corrective action, change, competence — these are leader standard work activities, and their outputs are the inputs to a higher-altitude one. Management review is not a separate species of meeting. It is the top-tier leader standard work event, running on an annual or semi-annual cadence, looking at defined inputs, producing a defined output of its own.
Which means management review inherits the quality of everything beneath it and cannot exceed it.
A management review fed by shaped, comparable, accumulated outputs can do analysis. It can see direction, compare periods, and make a resourcing decision with evidence attached. A management review fed by inconsistent lower outputs cannot trend anything, so it does the only thing available: it recounts the year from memory and anecdote, and produces minutes rather than decisions.
That is the honest explanation for why most management reviews feel like theatre. The meeting is not the problem. The meeting is downstream of a supply chain of outputs that were never designed to roll up.
The number of altitudes scales with the risk landscape
Two layers is the floor. The activity layer — risk, corrective action, change, competence — and management review above it. A small organization can run the whole system at that height and it will work, because one person can hold the connections in their head.
Larger and more complex operations break each area into subareas, and the same structure repeats at every level.
A worked version: departmental teams review corrective action trends weekly, at a frequency matched to their volume. Those outputs roll into monthly departmental risk evaluations. Those roll into a quarterly quality-level risk evaluation that identifies new and revised risks across departments. That rolls into management review.
Four altitudes, one repeating unit. Each level fires on a cadence and a trigger, looks at defined inputs — many of which are the outputs of the level below — and produces a shaped output for the level above.
The rule underneath: the more complex the risk landscape, the more nuanced the leader standard work. Tier count is not a maturity score to chase. It is a response to how much has to be watched and how finely it has to be divided to stay manageable.
And this is where the output argument stops being a preference. In a two-layer organization sloppy outputs survive, because the person assembling the review was in most of the conversations. Four altitudes up, an unshaped output at the bottom does not merely fail to trend — it breaks every rollup above it, and the failure surfaces at the top where it is hardest to trace. Height is what converts output shape from a nice-to-have into the load-bearing element. This is also why integrated management system consulting work spends so much time on output formats: with several standards in scope, the rollup paths multiply and the tolerance for improvisation drops.
Where this actually gets written down
The model needs an artifact, or it stays a good idea that each manager interprets differently.
On most implementations the artifact is a maintenance plan for the system — sometimes called a management plan. It is the document that specifies which activities exist, what altitude each one sits at, what fires it, what it looks at, what it produces, and where that output goes next.
Naming it matters because it locates the decision. Leader standard work is often treated as something individual managers design for their own areas, which guarantees that outputs will not be comparable across areas and will not roll up. The maintenance plan puts the design one level above the person performing the routine, which is the only place it can sit if the outputs are meant to connect.
It is also the document that makes the system survivable. A routine that lives in a manager’s habits leaves when the manager does. A routine specified in a maintenance plan is inherited. That is a large part of what distinguishes ongoing ISO 9001 maintenance from an annual scramble, and it is one of the first things an outsourced quality manager engagement puts in place, because an external party cannot run on institutional memory they do not have.
None of this is improvement work yet. It is the precondition for it. A continuous improvement framework needs a baseline and a direction of travel, and both of those are constructed from comparable outputs over time. Without them, improvement activity is a series of responses to whatever was most recently annoying.
The diagnostic
One activity, one afternoon.
Pick a leader standard work routine that actually runs — a review, a walk, an evaluation with a real cadence behind it.
Pull the last twelve outputs it produced. If you cannot locate twelve, that is the first finding.
Lay them side by side and ask whether they share a structure. Not whether they contain good information. Whether the same fields appear in the same places.
Try to answer one question that spans all twelve. Direction, recurrence, effect of the last change. If you have to read all twelve to answer it, the output was built as a record.
Name one other activity whose output should reach this one. Then check whether it ever has.
Follow the chain upward. Ask what of this activity’s output appeared in the last management review, and whether it arrived as data or as somebody’s recollection.
The findings usually sort into the two failure modes cleanly. Vertical: a full notebook nobody can aggregate. Lateral: separate notebooks that never meet. Most installs have both, and fixing the cadence — which is what usually gets proposed — addresses neither. Where the routines exist but the connections do not, that is ordinarily business process optimization work rather than a rebuild: the activities are already running, and what has to be designed is what comes out of them and where it goes.
The routine is not the deliverable. The routine is the mechanism.
What the routine produces is the thing the organization will still have in three years, and it will only have it if every instance was built to the same shape.
