A project can appear aligned while its stakeholders mean different things by success. One person may focus on delivery timing, another on user adoption, and another on maintaining an existing process. A summary stating that everyone wants a successful launch does little to expose those differences.
This guide creates an original success-definition matrix with ChatGPT. It keeps stakeholder statements separate from inferred criteria and proposed measures. Examples are fictional, and current documentation was checked on October 6, 2026. The matrix prepares a decision about success criteria; it does not manufacture agreement among the participants.
Gather statements with their context
Collect the approved brief, stakeholder notes, meeting records, and relevant messages. Give each statement a source ID, speaker or role, date, and project version. Preserve whether it describes a personal preference, a proposed criterion, or an already approved requirement.
Retain the exact condition that makes a statement meaningful. “Launch on time if the source review is complete” is different from an unconditional timing goal. Do not strip qualifications when normalizing the wording.
The ChatGPT usage reference covers comparing supplied information. Provide readable sources and confirm the relevant statements are present. A stakeholder not represented in the packet should remain outside the claimed comparison rather than receive an invented position.
Extract definitions before combining them
Ask ChatGPT to identify what each stakeholder explicitly calls successful and which conditions limit that statement. Keep references to benefits, measures, deadlines, and acceptance decisions separate. These concepts can be related without being interchangeable.
For example, “the guide should be easy to use” describes a desired quality. “At least the approved share of pilot participants completes the task” would describe a measurable criterion if the source actually supplies the threshold and task definition. The model should not invent that measure to make the first statement look more precise.
Record terms whose meaning is unclear, such as ready, adopted, complete, or reliable. The matrix should point to the definition gap and its practical effect. It should not choose a universal interpretation without the owner's review.
Build the success-definition matrix
Use one row per distinct stakeholder criterion or concern. A person may supply several criteria, and a single criterion may have support from several stakeholders. Preserve those relationships rather than force one summary row per person.
| Matrix field | What it records |
|---|---|
| Criterion ID | Stable reference for discussion |
| Stakeholder statement | Bounded account with source reference |
| Desired outcome | What the statement aims to achieve |
| Supplied measure | Actual metric or explicitly unstated measure |
| Threshold and timeframe | Conditions supplied by the source |
| Evidence source | Where progress or acceptance would be checked |
| Relationship | Shared, complementary, potentially conflicting, or unclear |
| Decision question | Clarification needed before approval |
Keep an owner field only when responsibility is established. The person expressing a goal is not necessarily responsible for measuring it or authorized to accept completion. Missing measurement ownership is a useful question for review.
Use a prompt that avoids false consensus
Supply the project boundary, statements, approved requirements, and matrix schema. Ask the model to compare definitions without creating a blended success statement as if it were already agreed.
Create a success-definition matrix from the supplied stakeholder statements.
Preserve source IDs, roles, dates, conditions, and approval status.
Separate desired outcomes from stated measures, thresholds, and acceptance authority.
Do not invent metrics, target values, deadlines, or stakeholder agreement.
Identify shared, complementary, conflicting, and ambiguous criteria with evidence.
Return decision questions and any stakeholder perspectives missing from the packet.
Label proposed measurement ideas separately from supplied requirements.Review every relationship label. Two goals can be complementary in principle but compete for resources in the actual plan. If the packet does not establish that tradeoff, mark it as a potential issue requiring investigation rather than a confirmed conflict.
Examine a fictional disagreement
Suppose a fictional workshop project has three statements. The delivery lead emphasizes running the workshop on the planned date. The learning lead emphasizes completing a reviewed activity pack first. The support lead emphasizes having a clear help route for participants.
The matrix can show three distinct criteria without implying that any stakeholder opposes the others. A possible conflict arises only if the activity review cannot finish before the planned date. The owner needs evidence about that dependency before deciding how to reconcile the goals.
A useful decision question asks whether the date is fixed or conditional on the reviewed pack. Another asks who confirms the help route is ready. A vague question such as “What does success mean?” may restart the whole conversation without resolving these specific differences.
Compare measurement scope and timing
Check whether stakeholders evaluate success at the same stage. Completion on launch day, useful adoption during a pilot, and sustained operation later are different horizons. A project can meet one and still need review against the others.
Preserve denominators and definitions for any supplied measure. A completion rate among registered participants differs from one among everyone invited. The matrix should not treat those as equivalent simply because both use a percentage.
If measures are absent, propose questions before proposed numbers. Ask what observable result would support the desired outcome and where that evidence could come from. Keep candidate measurement ideas separate until the owner approves the definition and threshold.
Prepare a criteria decision
Group the reviewed matrix into decisions the owner can make: confirm a shared criterion, clarify ambiguous wording, resolve a tradeoff, or request missing evidence. Present the source statements beside the decision questions so participants can check their representation.
Use the decision memo workflow if the criteria discussion requires an explicit choice between approaches. Keep any proposed combined definition labeled as a draft awaiting agreement.
Record the actual approved criteria with their scope, measures, timing, and acceptance role. Do not mark the matrix aligned merely because the wording has been made consistent. Agreement requires a real review decision, not a stylistic rewrite.
Maintain the criteria as the project changes
When scope or timing changes, identify the affected criteria and refer them for review. Preserve earlier versions so the team can understand which definition applied to a previous milestone. Avoid evaluating work against a criterion introduced after it was completed without acknowledging the change.
Track unresolved criteria questions explicitly. A missing measure may be acceptable for an exploratory stage if the owner records that decision. It should not disappear from the matrix simply because the project has moved forward.
Keep the source statements, reviewed matrix, decision record, and approved criteria together. The result makes expectations and their evidence visible while preserving the difference between a stakeholder's view, a proposed measure, and an agreed definition of success.
Frequently asked questions
Can ChatGPT create agreement from conflicting statements?
It can draft a comparison or a proposed definition. Stakeholders and the authorized owner must review and approve the actual criteria.
Should every goal have a numerical target?
Only when a suitable definition and threshold are established. Do not invent numbers to make a qualitative goal appear measured.
Are different goals always conflicts?
No. They can be complementary or apply at different stages. Label a conflict only when the evidence establishes an incompatible condition.
What if one stakeholder's view is missing?
Mark the perspective absent from the supplied packet. Do not infer that person's priorities or claim a complete comparison.
When should approved criteria be revisited?
Review affected criteria when project scope, timing, evidence, or acceptance responsibility changes, and retain the version history.
