Product guide / plain language
The system, without the jargon.
A compact guide to the objects and controls that make a Raison decision thread inspectable.
What it meanswhat changed
A signal describes the movement, its context and freshness. It is not a conclusion.
Inspect this section↓Quick answerThe essential answer stays immediate; every next section adds inspectable depth.
01Signal
A material change worth attention.
A signal describes the movement, its context and freshness. It is not a conclusion.
What must be namedName the exact metric or event, its unit, observation window and change against a dated baseline.
Why it mattersThe affected population, account tier and workflow stage prevent a broad average from hiding where the movement actually occurred.
How to inspect itShow the threshold, comparison period and exception rule that promoted this movement into a decision signal.
02Evidence
The inspectable source trail.
Evidence supports or challenges an interpretation and names what is missing.
What must be namedRecord the originating system, object or field, owner, refresh time and a route back to the source so a reviewer can reconstruct the claim.
How to inspect itExpose the formula, filters, joins and definition version between the source record and the displayed evidence.
What it does not proveName the missing source, period or population. The gap stays visible so confidence cannot grow simply because contrary evidence is absent.
03Decision
A human-owned next move.
A decision records alternatives, assumptions, approval, owner and outcome review.
What must be namedCompare the proposed action with alternatives and inaction, including trade-offs, reversibility and the signal each option should change.
How to inspect itRecord the named approver, date, conditions and scope. A recommendation remains non-operative until this gate is explicit.
Why it mattersAt the review date, compare the outcome with the original expectation and record continue, adapt or stop without rewriting the starting evidence.
DefinitionsSources, constraints, roles and review criteria—without marketing shortcuts.
04Hypothesis
A possible explanation remains separate from evidence.
It names supporting and conflicting observations, missing data and what could falsify it.
How to inspect itLink each supporting observation to a source and show why it increases plausibility without turning the hypothesis into a fact.
Why it mattersKeep counter-evidence beside supporting evidence so the team can see what weakens the proposed explanation.
What must be namedDefine the observation that would make the team reject or revise the hypothesis before acting on it.
05Scenario
A scenario compares an action with alternatives and inaction.
Assumptions, reversibility, owner, expected signal and guardrails make the proposal inspectable.
How to inspect itState the benefit, cost, affected users, reversibility and downside of each option in the same frame.
What it does not proveA reversible action lowers execution risk; it does not prove the underlying explanation is correct.
What must be namedAttach a measurable expected signal, owner and review date so the scenario can be evaluated after action.
06Confidence
Confidence explains evidence quality, not certainty.
It reflects source coverage, freshness, consistency and known gaps and must be paired with a rationale.
How to inspect itShow which relevant sources and populations are present and which remain outside the evidence set.
Why it mattersShow when each source last changed and the acceptable age for this decision; old evidence can be accurate but no longer useful.
What it does not proveList unresolved factors and how they could change the conclusion rather than hiding them behind a confidence label.
Continue the decision journey
See the complete product flow