Skip to main content

Decision Model

Purpose​

The decision model describes which management decisions are made along an initiative's path and what data is needed to make those decisions verifiable rather than a matter of taste.

It complements the decision framework: here the logic of choice is described, while the framework covers accountability, levels, and exceptions.

Principle​

Every initiative has four types of decisions:

  • promote — the initiative is ready for the next stage;
  • rework — there is potential, but data, an owner, a product, or artifacts are missing;
  • stop — the initiative is weak, risky, duplicate, or has no impact;
  • scale — the result is confirmed and applicable beyond a single process.

A decision must rely on the initiative card, tasks, checks, artifacts, and portfolio analytics.

Decision logic by stage​

StageMain questionPossible decisions
NewIs the idea worth examining further?move to assessment, clarify, merge, reject
AssessmentIs it worth allocating delivery resources?admit to delivery, change product, defer, reject
DeliveryIs the solution ready for impact verification?move to awaiting impact, rework, stop
Awaiting impactIs the result confirmed?close, support, scale, return to delivery
SupportIs continued maintenance needed?support, reconsider, replace, decommission

Data for the decision​

Minimum set:

  • a description of the business problem;
  • the expected impact;
  • the initiative owner;
  • the AI product;
  • the result of the similar-initiative check;
  • priority;
  • data and security constraints;
  • tasks and delivery status;
  • the impact review date;
  • the actual result after adoption.

If the data is insufficient, the correct decision is not to "let it move forward", but to return the initiative for clarification.

Decision quality​

A good decision:

  • is understandable to any participant a month after it is made;
  • has criteria;
  • is saved in the initiative's history;
  • names an accountable owner;
  • affects the next step;
  • does not require guessing why the initiative ended up at the current stage.

A bad decision:

  • is made verbally;
  • has no reason;
  • is not linked to the card;
  • depends only on the initiator's status;
  • leaves no trail in the system.

Connection to the AI assistant​

The AI assistant can prepare recommendations:

  • which fields are not filled in;
  • why an initiative is similar to another;
  • which product might fit;
  • which risks should be checked;
  • what is blocking the next transition;
  • how to phrase a decision for a stage gate.

But the final decision is made by a person or a designated body. The AI assistant must not substitute for the decision owner.