Six months after a budget change, the business can usually find the amount.
Finding the reason is harder.
The explanation may be scattered across a presentation, a message thread, an analyst's notebook, and the memory of the person who approved it. By the next planning cycle, the original uncertainty has often disappeared. The eventual result starts to look inevitable.
That makes it difficult to learn.
The MIZ OKI Decision Graph is designed to keep the decision together. What was known at the time? Which explanations were considered? What alternatives were rejected? Which policy applied? Who authorized the move? What outcome did the team expect?
The realized result belongs beside those answers.
This is an operating model built on connected records. It can work with the warehouse, the metric definitions, and the systems that hold the underlying evidence. Those systems remain useful. The decision record gives their information a shared purpose.
Time is essential to that record. A late-arriving refund might change how we understand a campaign today. It should not silently rewrite what the approver knew on the day of the decision. Keeping both views allows a fair review of the action and the evidence process.
MIZ OKI organizes that work through seven stages, called SRPVDAL:
- Sense: identify a meaningful change and retain its source and context.
- Reason: compare explanations and keep the evidence against them.
- Plan: set out alternatives, including leaving the current position unchanged.
- Validate: check the evidence, economics, policy, and operating constraints.
- Decide: establish the named person or delegated authority responsible for the choice.
- Act: carry out only the scoped action that has been authorized, when the execution path is enabled.
- Learn: compare the expected outcome with what actually happened and preserve the lesson.
The sequence is useful because each stage answers a different question. An explanation does not select an intervention. A selected intervention does not grant authority. An action that ran successfully does not prove that it created business value.
Consider an illustrative rise in acquisition cost. The initial suspects might include creative fatigue, a tracking problem, a landing-page fault, and inventory pressure. The record should preserve the competing explanations as they are tested. A team reviewing the incident later needs to know why it acted on one and set the others aside.
The same applies to a hold. If the team waits for better cost data, that choice deserves a reason and a review condition.
Learning also requires care. An improvement after a change is an observed outcome. Where the team claims that the intervention caused the improvement, it needs an appropriate causal comparison. The graph should preserve that distinction as carefully as it preserves the approval.
A shared record reduces dependence on one person's recollection. Finance can examine the economics. Operations can inspect the constraints. A new team member can understand a decision without reconstructing a meeting they never attended.
That is the institutional memory I want the product to build: evidence about how the business makes choices, including where its judgment was wrong. The Media product page labels the broader capability to learn which interventions work across conditions as Roadmap. Preserving a record and establishing that learning capability are separate steps.
The Decision Studio outcome view illustrates this record with a fictional scenario. The demonstration keeps temporary session state and offers an export of its decision record.
Walk through the Decision Graph, then explore the seven-stage operating loop.