Skip to main content

// Process · Decision path

From ambiguity
to shipped
product

The work is not a fixed checklist. The right methods make risks visible, align the team and connect insight to implementation.

01 / The decision path

Six stages, one route

Each stage answers a question, leaves an artifact behind and hands something specific to the next. Methods are chosen per project: not every project used every stage or artifact.

  1. Illustrative artifact · fictional example

    01 / 06 · Business goals, user needs, constraints

    Understand

    What job is actually being done, by whom, and under which constraints?

    Artifact
    Discovery notes, a stakeholder map and a list of constraints that cannot move.
    Decision
    Which problem is worth solving first, and what is out of scope.
    Failure state
    Designing for the ideal process instead of the one people run. A missing role or rule surfaces late.
    Handoff
    A problem statement and constraints that product and engineering agree on.
    In the workStatus: Shipped client workDoD workflow modernization Discovery started with the existing process: how information moved and where correspondence stalled.
  2. Illustrative artifact · fictional example

    02 / 06 · Workflows, information architecture, journeys

    Model

    How does work move, where does it wait, and who owns each step?

    Artifact
    A workflow map, the information architecture and a journey across roles.
    Decision
    One visible owner, status and next step for every request.
    Failure state
    Hidden handoffs: a request waits with nobody visibly responsible for it.
    Handoff
    A state model engineering can check against data and permissions.
    In the workStatus: NDA-safe reconstructionOrbit operations Every request shows its owner, status and next step.
  3. Illustrative artifact · fictional example

    03 / 06 · Wireframes, prototypes, alternatives

    Explore

    Which interaction model solves it, and what does each option give up?

    Artifact
    Wireframes and prototypes, with alternatives compared side by side.
    Decision
    The direction to take forward, with the rejected options and reasons written down.
    Failure state
    Polishing the first idea before any alternative has been compared.
    Handoff
    A chosen direction and a prototype the team can react to.
    In the workStatus: NDA-safe reconstructionVector readiness Compare periods, spot the outlier, follow it to the source.
  4. Illustrative artifact · fictional example

    04 / 06 · Components, patterns, design systems

    Systemize

    What should be reusable, and what is truly a one-off?

    Artifact
    Tokens, components and patterns, with every state documented.
    Decision
    Status, empty, error and permission states are specified once and reused.
    Failure state
    Each screen invents its own status colors and design debt grows quietly.
    Handoff
    Component specs and states ready for code and Storybook.
    In the workStatus: Shipped client workDoD workflow modernization Program leaders designated the UI/UX direction and reusable patterns as the standard for future applications in that program.
  5. Illustrative artifact · fictional example

    05 / 06 · Engineering partnership, design QA, iteration

    Ship

    Does the build match the intent, in every state and at every screen size?

    Artifact
    Specifications, acceptance checks and design QA notes.
    Decision
    What ships in this increment and what waits for the next.
    Failure state
    Treating handoff as the end, so edge states are found in production.
    Handoff
    A release with QA findings resolved or logged for later.
    In the workStatus: Illustrative design artifactDesign to code Figma kits, specs and the components they become.
  6. Illustrative artifact · fictional example

    06 / 06 · Feedback, measurement, next iteration

    Learn

    Is it working, and what should the next increment change?

    Artifact
    A feedback log, candidate measures and a brief for the next increment.
    Decision
    Which signal is strong enough to change the plan.
    Failure state
    Measuring what is easy instead of what matters, or claiming results nobody measured.
    Handoff
    A prioritized next increment.
    In the workStatus: Shipped client workDoD workflow modernization Processing time: reported as 1–2 months before, observed in dashboards at 1–2 weeks after. A qualified comparison, not a controlled study.

02 / The lens

Insight → decision → design response

  1. 01

    Insight

    What did we learn about the job, the system or the constraint?

    Fictional Orbit exampleReviewers find missing evidence only after they open a request.

  2. 02

    Decision

    Which direction best solves the problem, and what does it sacrifice?

    Fictional Orbit exampleEvidence completeness matters more than minimizing clicks.

  3. 03

    Design response

    How does the interface and interaction model express that decision?

    Fictional Orbit exampleKeep the record and queue together, explain the blocker, give the requester a way to recover.

03 / Representative Orbit delivery brief · fictional example

One decision. Shared across the team.

A reviewer finds missing evidence after opening a request. The proposed response keeps the record and queue context together, explains the blocker, and gives the requester a recoverable path. This brief illustrates a collaboration method; it is separate from my résumé-reported delivery history.

Shared decisionEvidence completeness before speed

Product manager / project lead

Agree the user problem, prioritize the recovery path, define the scope of the first increment, and defer bulk approval until individual review is clear.

Engineering

Review the state model and API constraints. Keep record identity stable, preserve entered feedback on failure, and enforce permission rules in the service.

QA

Turn state specifications into acceptance checks: missing evidence blocks approval; failure preserves input; retry avoids duplicate submission; narrow screens and keyboard use retain access.

Business leadership / operational stakeholders

Review the approval responsibility and audit requirements. Agree that evidence completeness matters more than minimizing clicks, and select measures for a later evaluation.

  1. 1. Shape an increment. Model the workflow and compare low-fidelity options with product and engineering before polishing the UI.
  2. 2. Specify and build together. Review component states and acceptance criteria, then implement the smallest coherent flow.
  3. 3. Inspect and iterate. Compare the build with design intent, resolve QA findings, and use feedback to plan the next increment. This demonstrates iterative delivery without claiming a particular Scrum ceremony or measured result.
Inspect the supporting interaction contract

04 / Design + engineering

Design does not end at handoff

My front-end background helps me evaluate constraints early, review built experiences against design intent, and work with engineering to resolve issues before release.

Design tokenStatus / warning
ComponentNeeds review
Implementation concept<Status variant="warning" />
  1. 01Figma
  2. 02Specification
  3. 03Implementation
  4. 04Design QA
  5. 05Refinement
  6. 06Release

05 / Outcomes and evaluation

How I think about success

I define desired outcomes early, then choose measures with the team that can show whether the experience is working.

  • Efficiency

    Task completion time · steps required · manual work removed

    Set with the team
  • Quality

    Error rate · completion rate · exception handling

    Set with the team
  • Adoption

    Feature usage · repeat usage · workflow completion

    Set with the team
  • Experience

    User feedback · usability · support friction

    Set with the team
  • Delivery

    Component reuse · implementation consistency · design debt

    Set with the team

These are candidate measures for future evaluation, not reported project results. Not every engagement exposes every metric, but I define desired outcomes early so decisions can be evaluated against something more meaningful than preference.

06 / How products behave

Interaction design

A polished interface is more than its resting state. I design how products respond, communicate status, preserve context, and recover as people move through complex workflows.

Want the full set? The interaction lab has twelve working patterns.

07 / The toolkit, used when it answers a real question

  • Discovery workshops
  • Interviews
  • Workflow maps
  • Information architecture
  • Prototypes
  • Usability feedback
  • Component specifications
  • Design QA

I use discovery workshops, interviews, workflow maps, information architecture, prototypes, usability feedback, component specifications, and design QA where they help answer a real product question. I avoid artifacts that exist only to decorate a case study.

See the method in the work.