Question and scope
This protocol measures how often a cross-team review takes place without the person who has authority to make the decision under discussion. A review qualifies when it is a recurring or scheduled meeting where two or more teams examine a shared item and expect a decision, a date, or an approval. The decision-maker is the single person whose agreement makes that outcome binding. The result is a share of reviews held without that person, a count, and a list of decisions deferred as a consequence.
The measure is about authority in the room, not attendance in general. A review can have full attendance and still lack the person who can approve the change. It can also have a sparse room where the one decision-maker settles the matter in ten minutes. Counting heads without identifying authority produces a comforting number that does not predict whether the meeting produced anything usable.
Scope is one organization, a defined set of cross-team reviews, and an observation window of at least four weeks. Include standing reviews, stage gates, and approval meetings that recur on a calendar. Exclude one-off kickoffs, informational briefings, and meetings whose only purpose is status sharing. Name the meeting list and the window so a second analyst can rebuild the population.
Why attendance is the wrong proxy
Most teams already track whether the right departments showed up. That metric fails this question because departments do not decide; people do. A delivery manager may attend every review and still route an exception to a director who never joins. The room looks complete on the attendance sheet while the actual decision waits for a separate conversation.
Attendance also hides the difference between a delegate and a proxy. A delegate can speak for a team but cannot commit it, while a proxy can commit but may not understand the item. Neither substitutes for the person named in the approval rule. The protocol therefore records who attended and separately records whether the approval authority was present.
Deferred decisions have a cost that attendance metrics cannot see. A review that ends with "we will take it offline" adds a day or a week to every affected item. Repeated across a quarter, those deferrals become the reason dates slip. The share measured here makes that pattern visible before it turns into an excuse.
Defining the decision-maker
Define the decision-maker from the written approval rule, not from seniority or habit. The rule may live in a delegation-of-authority document, a procurement threshold, a change policy, or an escalation matrix. It should name a role or a person for the decision type under review. If no rule exists, the team must name one before the measurement starts, or the result will reflect opinion rather than authority.
Record the decision type for each review because authority varies by type. A vendor renewal above a set value may need finance approval, while a scheduling change may need only the delivery lead. The same meeting can require different decision-makers for different agenda items. Treat an item as uncovered when its specific decision-maker is absent, even if another senior person is in the room.
Be careful with shared authority. Some rules require two signatures or a committee quorum, in which case the decision-maker is the full set and the review counts as covered only when that set is present. Record the rule explicitly so coding is reproducible. Ambiguity here will produce disagreements that no amount of recoding can settle.
Unit of analysis and exclusions
The unit is one review occurrence with one agenda and one decision requirement. A recurring weekly review produces one unit per occurrence, not one unit for the series. Splitting by occurrence shows whether absences cluster around quarter end or holidays. It also prevents one well-attended session from standing in for a quarter of missed decisions.
Exclude occurrences that were cancelled, because a cancelled review did not run without the decision-maker; it did not run at all. Exclude sessions explicitly convened as preparation with no decision expected. Include sessions where a decision was expected but the item was withdrawn, and code the withdrawal reason separately. Count remote presence as presence when the person could hear the item and decide.
Record the agenda as published before the meeting rather than the version circulated afterward. A retroactively edited agenda can hide that a decision item was dropped. Keep the original invite and attachment as evidence. If the agenda changed during the meeting, note the change and code the item actually discussed.
Building the review inventory
Start with the shared calendar and list every recurring cross-team meeting in the window. Add stage gates and approval sessions that appear only in project plans. Ask each team lead which reviews they attend that involve another team, because some sessions are scheduled by one group and invisible to the others. The inventory should cover the reviews where decisions happen, not only the ones with formal names.
For each meeting, record the owner, cadence, invited roles, and decision types on the standing agenda. Mark meetings where the decision-maker is not on the invite at all, since those are structurally unable to decide. That flag is different from an absence and should be reported separately. A meeting that never invites the approver will fail every week for a reason the team can fix once.
Reading attendance and authority
For each occurrence, record the invitees, the attendees, and the role each attendee holds. Then compare the attendee list with the decision rule for each agenda item. Code the item as covered when the required decision-maker was present for the discussion, and uncovered when the person was absent, left early, or joined after it concluded. Presence for the whole meeting is not required; presence for the item is.
Capture the outcome as well. Record whether a decision was made, deferred, or referred elsewhere. A covered item can still be deferred for good reasons, such as missing data. An uncovered item that somehow produced a decision is worth flagging, because it may mean the written rule is out of date. Both conditions are useful and should not collapse into one pass or fail.
Evidence standards
Accept attendance evidence from the meeting record, the calendar response list, or the minutes, in that order of preference. A verbal recollection is the weakest source and should be used only when no record exists, with the source noted. Do not infer attendance from a name in the minutes unless the minutes distinguish attendees from people mentioned. Minutes often list everyone discussed and only some people present.
Accept the authority map from a written rule, and record the document and version. If the team operates from an informal understanding, write that understanding down and have the responsible manager confirm it before coding begins. An unconfirmed authority map makes the whole measure provisional. Report that status in the result rather than presenting a precise share built on guesswork.
Do not add the decision-maker to an invite during the audit and then count earlier occurrences as covered. That changes the population instead of measuring it. If the team fixes the invite while the measurement runs, record the change date and report pre-change and post-change results separately. The original state is what the headline describes, and the change is what the team did about it.
Sampling and verification
Use a census when the number of occurrences is small enough to review, which is common for a single program with four to six recurring reviews. Above roughly two hundred occurrences, draw a random sample and publish the method. Stratify by meeting and by week so a busy period does not dominate. Freeze the sample before reading attendance records.
Have a second reviewer code at least ten occurrences independently and compare the results. Resolve disagreements by returning to the written rule and the original invite, not by debating who usually decides. Record the resolution so later reviewers apply the same interpretation. If disagreement exceeds one in five items, rewrite the definitions and recode before reporting.
Decision rules before coding
Decide whether a standing delegate counts as the decision-maker when the rule permits delegation. If the rule names a delegate in writing, count that person; if the rule is silent, do not. Decide whether a decision-maker who joins by phone or video counts as present. The practical answer is yes when the person can hear the item and state a decision, because location is not the constraint being measured.
Decide how to treat an item resolved before the meeting by email. Code it as not requiring a decision in the room, and exclude it from the denominator for that occurrence. Do not count it as covered, because the meeting did not do the work. Keep a separate count of pre-resolved items so the report shows how much of the agenda is settled before anyone arrives.
Analysis and reporting
Report the share of decision items reviewed without the required decision-maker present, with raw counts beside every percentage. Add the share of meetings that never invite the approver, since that is a structural defect rather than a scheduling accident. Report deferral counts and median deferral length alongside coverage. A coverage share without the consequence of missing coverage is easy to dismiss.
Break the result down by meeting, decision type, and week. Suppress groups with fewer than five items so one unusual session does not read as a pattern. Look for clustering: if absences concentrate in one meeting, the fix is a calendar change, while absences spread across all meetings point to a delegation rule. Those are different problems with different owners.
Pair the number with the decisions that matter most. A quarterly vendor review that cannot approve a renewal costs more than a weekly coordination call that cannot approve a date change. Weighting by consequence is a judgment, so present it as a separate view rather than folding it into the headline. The weekly blocked-work review is a natural place to discuss high-consequence items.
Limits and sources of error
The sources cited here describe internal control and project governance in general terms; they do not provide a benchmark for decision-maker attendance. A low coverage share may mean the authority map is stricter than practice, or that decisions are routinely made outside the meeting. Neither possibility is visible from the attendance record alone. Treat the share as a prompt to inspect the approval rule, not as proof of poor governance.
This measure cannot see decisions made in hallways, chat threads, or side calls. If a large share of items are resolved before the review, the meeting may be a ratification step rather than a decision forum. Check a sample of pre-resolved items to understand that pattern and report it as a limitation. A precise share built on the wrong model of the meeting is worse than an honest description of how decisions happen.
Review cadence
Review the result at a fixed point and end with one change, one owner, and one check date. A practical first change is to add the named approver to the invite for the specific agenda item rather than the whole meeting. Repeat the measurement after two cycles before setting a target, because one period may reflect holidays or a single large escalation. Keep the authority map with the result so the next reviewer can compare like periods.
Retire the measure when every decision item has its required approver present or formally delegated, and the share stops changing. If the team keeps it, publish the structural gap and the absence gap separately, because they have different fixes. Link the result to the decision review protocol when a decision needs a future checkpoint. The purpose is a review that can decide, not a calendar that looks full.
Sources
- U.S. Government Accountability Office, Standards for Internal Control in the Federal Government, reviewed November 17, 2026.
- Project Management Institute, PMBOK Guide, reviewed November 17, 2026.
- National Institute of Standards and Technology, SP 800-39: Managing Information Security Risk, reviewed November 17, 2026.
