Fewer forums, stronger authority to decide

Light governance addresses a familiar tension: teams need to act quickly, while leaders need to explain the commitments they approve. Removing a meeting does not resolve that tension. A decision may keep circulating by email between several managers, with nobody authorised to close it. Its cost disappears from the calendar but remains in waiting time, repeated work and implicit commitments. Useful simplicity brings authority closer to the work, specifies its limits and retains a record proportionate to the consequences.

This retrospective examines choices available in 2023 and was written in 2026. All situations, amounts and timings below are fictional teaching examples, not accounts of CYTIZEN engagements. The central proposal concerns organisation: delegate reversible local choices, name an arbiter for cross-functional choices, and preserve independent assurance where exposure requires it. The level of authority follows impact and reversibility rather than the personal preferences of a requester or the prominence of a project.

In practice: the file that keeps returning

Imagine a fictional distributor considering a replenishment rule. Logistics sees excess inventory. Sales fears shortages. IT can configure the rule but has no mandate to accept a lower customer service level. Three forums examine the same proposal, each asking for another metric. Meanwhile, the existing rule keeps placing orders. The status quo is an effective decision, even though nobody signed it. The first useful intervention is to make this consequence visible and identify who can accept the trade-off.

Review ten recently blocked decisions with operational staff. Record the question, the date when minimum evidence became available, the authority sought, the number of reviews and the outcome. Project opening dates do not substitute for the date a decision became ready. Distinguish missing information, a substantive disagreement, an insufficient delegation and an unavailable decision maker. These problems need different remedies. Adding one committee to all four merely conceals the diagnosis under a common procedure.

Delegation with an explicit boundary

In this example, the operations director delegates changes affecting one warehouse, reversible within a day and leaving customer commitments unchanged. Logistics decides inside that boundary. Changes affecting several warehouses or delivery promises go to a named arbiter. A documented security objection blocks execution until resolved; an unsupported preference remains advice. Publish these rules beside the decision register so that new staff can see both the authority and the conditions attached to it.

Fictional questionAuthorityEvidenceBoundary and outcome
Lower one warehouse thresholdLogistics leadStock, shortages, tested reversalReversible trial; stop at agreed shortage limit
Change delivery promisesOperations directorCustomer and margin impact, sales adviceExplicit cost versus service choice
Remove a compliance checkAuthorised owner with compliance expertApplicable requirement and replacement controlNo implied delegation; retain until evidence exists

This is a completed model, not a universal authority schedule. Money alone cannot define its boundaries: a small access change may expose sensitive records, while a larger expenditure may be fully reversible. Combine cost, reach, criticality and reversibility. One accepted example and one refused example often make a delegation clearer than several abstract statements.

One decision owner, several contributors

The decision owner frames options, requests the necessary advice and signs the conclusion. This person need not prepare every calculation. The business owns service consequences; IT explains implementation constraints; finance checks the economic reasoning; relevant specialists assess risk. The facilitator arranges the discussion and maintains the record but does not become the arbiter by default. This prevents authority from falling to the loudest voice or to whoever happens to write the minutes.

Consultation does not create a general veto. Specify whether advice is mandatory or advisory and when it is due. Silence is not approval of an unexamined risk. For a low-risk local choice, the arbiter might proceed while recording missing advice and the accepted limitation. Where implementation depends on a statutory authorisation, that authorisation must be obtained. Speed comes from distinguishing these situations, rather than treating all silence as permission.

A short record that supports execution

Record the question, genuine options, decisive assumptions, authority, conclusion, effective date and review trigger. Briefly explain why the alternative was rejected and link the underlying evidence. If an assumption fails, reopen the decision against its trigger. Do not restart every discussion merely because a new participant arrives. The record should help staff execute and learn, rather than become an archive maintained only for appearances.

Measure waiting and decision quality together

Decision lead time equals conclusion date minus the date minimum evidence became available. The median describes typical cases; the ninetieth percentile reveals neglected files. The rework rate is decisions reopened for inadequate preparation divided by decisions concluded. Reopening because genuinely new information emerged belongs in a separate category. These distinctions discourage improvement achieved by declaring files ready late or hiding corrections.

In a fictional dataset, twelve ready files wait between one and nine days, and three are returned because they lack an owner. Appoint owners before demanding a universal twenty-four-hour response. Participant hours per concluded decision can reveal governance load, but a reduction is useful only if execution follows and avoidable incidents do not rise. Fast conclusions based on poor understanding save meeting time while increasing operating costs.

When to remove or suspend a forum

A forum earns its place through decisions that no existing role can conclude. Information-only meetings may become an accessible written update. Duplicate authority should move to the place where accountable people can inspect consequences. Independent assurance must survive even if its meeting format changes. In the example, a forum closes after three sessions without a distinct decision, provided its information and control duties are explicitly transferred to an owner.

Correct or suspend the arrangement if delegation permits commitments beyond the mandate, risk alerts have no recipient or conclusions are not executed. A decision without an action owner and an observable first effect is unfinished. Keep its original reasoning when checking execution; rewriting the record after the outcome becomes known destroys its value for learning.

Adjust authority to sector constraints

In manufacturing, production scheduling might be delegated while equipment safety changes require qualified approval and verification. In banking or insurance, a commercial choice does not replace the examination of controls or customer commitments. In public services, a journey owner must consider legal rules, accessibility and appeal routes. Appropriate delegation depends on what can be lost and how a failure can be repaired.

Multisite operations add displaced consequences. A warehouse reducing inventory may increase another site's urgent transport costs. Include the units absorbing those costs in the decision boundary. Ask where the benefit appears and where the burden lands. If they differ, a cross-functional arbiter may remain necessary. Light governance makes interdependence visible when relevant rather than pretending it has disappeared.

Start with a recurring decision family and write a one-page mandate. Test two ordinary cases and one boundary case, then observe subsequent files. Staff should be able to challenge the rule using a documented example. Expand an overly narrow delegation or strengthen insufficient assurance. The practical test is whether someone knows where to take a difficult question and receives an executable conclusion without creating another route.

Sources and method

This is an editorial analysis, not a standard or a claim about client results. Two official references were checked in October 2026. Their general principles inform the method; thresholds and stopping conditions remain teaching proposals.