A journey is a state machine, not a funnel picture
Answers Mereology · Topology · Modality · Semantics
A Journey is the machine; a Journey State is a position inside exactly one Journey and is typed by exactly one Consumer State; a Journey Transition belongs to one Journey and names exactly one from-state and one to-state. A Journey begins at exactly one of its own positions. An Event is what triggers a transition, every transition names at least one, and only an event resolvable to one person may trigger a move. What a move asks of activation is not stored on the move: a campaign targets the condition it acts on, and the question areas that read the journey advise the purposes whose campaigns act on it. A stage list with no legal transitions is a picture, not a model.
What goes wrong without it
A funnel diagram names stages and says nothing about how anyone moves between them, so it cannot be validated, queried or executed. A state machine names the legal moves and what triggers each, which is the difference between a journey a system can run and a journey a deck can show.
What proves it — 3
Each is a gate the build runs: a check inside a ring, a whole ring, or an eval beside them. A build stops at the first that refuses.
- registry integritytyped relationships
Every relationship has declared endpoint kinds and a governed cardinality. - registry integritydistinct definitions
No two rows of different kinds are defined in the same words, and no definition restates a field the row already carries, so a distinction the model commits to is carried where a reader meets it and nothing the model knows is stored in two places at once. - registry integrityjourney integrity
Every journey begins at one of its own positions and carries at least one legal move, its positions sit on a contiguous run of rungs, and every position either has a move out or realises a condition another journey leaves from, so no journey is a picture and no position is a dead end.
Derived from — 3
- W3CState Chart XML (SCXML): State Machine Notation for Control Abstraction
- Kimball GroupDeclaring the Grain
- David Harel, Science of Computer Programming 8(3), 1987Statecharts: a visual formalism for complex systems
Kinds it licenses
Relationships it licenses
begins at · belongs to journey · realizes consumer state · from state · to state · triggered by
Where this model stands
Transitions carry exactly one from-state and one to-state, so the journey is walkable in both directions without a second asserted fact.
Consumer State stays a governed condition and Journey State stays a position inside one journey; neither is collapsed into the other.
Journeys meet where two of their positions realise the same condition: a person who is committed stands at the end of acquisition and the start of onboarding, and an account holder at the end of onboarding, the settled position of retention, and the start of advocacy. So the lifecycle reads as one walk from no relationship to bringing others in, without any journey standing in for the ladder.
A journey declares the position it begins at, so where a walk starts is stated once rather than inferred from which position nothing leads to.
Every move names at least one event that triggers it, including the fact that nothing happened within a window and the fact that a model scored someone, so no move is a picture and no machine document has to guess.
A rung is a superstate over the conditions that sit in it, and journeys are orthogonal regions that share conditions, which is the hierarchy and concurrency a statechart adds to a flat machine, and why a position realises a condition rather than being one.
An arrival is a move from the position the journey begins at, triggered by the fact that first made the person known. How they arrived is read from the surface that fact was captured on and the channel the surface sits in, and is not stored a second time as a classification of the surface or the move: a vocabulary that once did so was inherited from another domain's way of classifying how its services are found, and it is gone.
What it does not claim
Several observed events may legitimately trigger the same move, so a move names one or more triggers rather than exactly one; narrowing that to a single triggering event per move needs a source that proves one.
RUL-011