Question and scope

This protocol measures how often a cross-team decision record names the alternative that was rejected and the reason it lost. A decision record that lists only the chosen option looks complete but cannot support a later review. When the same question returns, the team reopens the debate because no one can explain why the other path was set aside. The measure tests whether the record preserves the choice, not just the outcome.

Scope is one decision type handled by one coordination team during a defined window, such as all decisions recorded in the last two quarters. Include decisions made in meetings, tickets, and written approvals. Exclude routine scheduling choices that carry no lasting trade-off. State the decision type, the system of record, and the window so the boundary can be reproduced; a narrow scope keeps the coding effort manageable.

What counts as a recorded alternative

An alternative is a specific option that was seriously considered and could have been chosen. A vague note that says other options were reviewed does not qualify. The record should name the option, the main reason it lost, and the condition that would make it viable later. Those three elements turn a rejected path into a usable reference rather than a footnote.

The chosen option does not need a long justification. A short statement of the constraint that decided the matter is enough, such as a fixed date, a budget limit, or a dependency that could not move. The point is to preserve the reasoning while it is still fresh. A record written months later will reflect the outcome, not the information available at the time.

Unit of analysis and exclusions

The unit is one decision with one decision maker, one chosen option, and one recorded rationale. Split a bundled decision when each part could have been decided separately. If the parts share a single constraint and decision maker, keep one unit and record that the scope was shared. Use the record as it existed at the decision date, not a version edited during the audit.

Exclude decisions that were reversed before any work started, because the final record may describe the second choice only. Exclude duplicate entries created by two systems for the same decision, and measure a reversal separately if it matters. Report the exclusion count and a short reason for each excluded unit. The denominator should be clear enough that another reviewer would reach the same population.

Start and stop clocks

Start the clock at the decision date recorded in the source, not the date the record was last edited. If the decision date is missing, use the first message that states the choice and mark the date as estimated. An estimated date is weaker evidence and should be reported separately. Do not use the date the record was moved into a wiki or a decision log.

Stop the clock at the first review of the decision or at the end of the observation window, whichever comes first. Record whether the rejected alternative was reconsidered at that review. A decision that was never reviewed is not a failure by itself, but it limits what the record can tell the team. Report the share of units with no review as a separate figure.

Evidence and data dictionary

Each unit needs an identifier, the decision question, the chosen option, the rejected option, the deciding constraint, the decision maker, and the decision date. Add the review date, the reviewer, and whether the choice was reopened. Link to the message, ticket, or minutes where the decision was made. The evidence should let a reviewer reconstruct the trade-off without interviewing the participants.

Define the terms before coding begins. Distinguish a rejected alternative, a deferred alternative, and an option that was never viable. Decide whether an alternative that lost only on cost counts as recorded when no cost figure appears. Record one example of each category in the dictionary. A shared vocabulary keeps two reviewers from counting the same note differently, which protects the headline share.

Sampling and test design

Use a census when the window contains fewer than roughly one hundred decisions. Above that volume, draw a random sample from the decision log and document the method. Stratify by decision type or by the team that owned the choice, so one high-volume category does not dominate. Freeze the sample before reading the rationale to reduce selection bias, and keep the frozen list beside the coding sheet.

Have a second reviewer recode at least ten decisions and compare the results. Resolve disagreements by returning to the source record rather than by discussion, and record each resolution. If the disagreement rate is high, revise the definition of a recorded alternative and recode the full sample. A measure that depends on one reviewer's judgement will not survive a change of reviewer.

Decision rules before coding

Decide how to treat a decision where the rejected option appears only in a comment thread. One option is to count it as recorded because the evidence exists. Another is to count it only when the main decision record names it. The choice changes the headline share, so write it down before the first row is coded and keep it with the coding sheet.

Decide how to handle a decision that had no realistic alternative, such as a legal requirement with one compliant path. Count it as not applicable rather than as a failure, and report the not-applicable share separately. Also decide whether a deferred option that later became the chosen one counts as recorded. Both choices are defensible if the report states which rule was applied.

Analysis and reporting

Report the share of decisions that name a rejected alternative, the share that explain why it lost, and the share that state a reopening condition. Add the share reviewed within the observation window and the share reopened at review. Show the raw count beside every percentage, especially when the sample is small. A share based on six decisions should be presented as a count, not a rate.

Break the result down by decision type, decision maker group, and system of record. Suppress groups with fewer than five units so one unusual decision does not look like a pattern. Pair the completeness result with the time to decide, because a fast decision with a thin record may create more rework later. The combination is more informative than either figure alone, and the weekly blocked-work review is a practical place to test whether the record survives contact with the next decision.

Limits and sources of error

The cited public sources describe appraisal, internal control, and iterative delivery in general terms; they do not provide a benchmark for a specific decision log. A missing alternative may mean the choice was obvious, or that the recorder was rushed. The measure also depends on whether the log preserves the original entry. A system that overwrites the rationale will understate completeness.

Backfilled records and reconstructed rationales are common sources of error. Check a sample against the original messages and mark any entry written after the decision. Exclude a unit when the decision date or the chosen option cannot be verified, but report the exclusion count. A precise number built on guesses is worse than a range with a clear caveat, especially when the log is maintained after the fact.

Review cadence

Review the result at a fixed point and end with one change, one owner, and one check date. Repeat the test after two full cycles before setting a target, because a single quarter may reflect an unusual mix of decisions. Keep the definition and the sample method with the result so the next reviewer can compare like periods. Mark any template change on the same timeline as the data.

Retire the measure if it stops informing a decision. If the team keeps it, publish the rejected-alternative share and the reopening share together, because a complete record that no one reads does not improve decisions. The purpose is a record that helps the next team choose faster, not a longer document. Stop when the team can show that pattern consistently across several cycles, and compare the result with the exception aging test before changing the template.

Sources