← Glossary
State model
A state model is the backbone of a governable experience. If you can’t describe the states, you can’t reliably design the UI, instrument outcomes, or recover from failure—especially in multi-step services and agentic workflows.
Definition
- A state model defines the finite set of meaningful states a journey can be in, and the transitions between them.
- It includes triggers, owners, timeouts, failure states, and recovery paths.
- State models are used by design, engineering, operations, and support—because they describe the same reality in one language.
Why it matters
- State models eliminate ambiguity: everyone knows what “pending” means and what can happen next.
- They prevent edge cases from becoming production incidents and support escalations.
- In agentic systems, the state model is the control plane: it governs reversibility, handoff, and audit trails.
- State models are also communication tools: they make cross‑functional debates concrete (“which state are we in?”).
- In practice, this is where many digital programs fail: the concept is understood, but the operating discipline is missing.
Common failure modes
- Only happy path states: everything else becomes “error”.
- UI states that don’t map to backend states: mismatched truth creates user confusion and broken support.
- No ownership: who transitions the state—user, system, operator, partner—remains unclear.
- No time dimension: pending states without SLAs or expiry behaviors.
- States are documented once and never maintained as the product evolves.
How I design it
- Start with outcomes: what does “done” mean? Then work backward through required states.
- Model failure states deliberately: retriable, blocked, rolled back, manual takeover.
- Define visibility rules: which states are shown to users vs operators, and with what cues.
- Design transitions with guardrails: idempotency, confirmation patterns, and escalation hooks.
- Use the model as a governance artefact: changes require decision ownership and versioning.
- Review state models in critique rituals and quality gates; treat them as release-critical for high-stakes flows.
- Treat it as a repeatable pattern: define it, test it in production, measure it, and evolve it with evidence.
Related work
Proof map claims
Case studies
See also
Contact
Let’s discuss a leadership role, advisory work, or a complex product challenge.