Work

Giving every design a clear status, and every decision a clear owner

Context

A rough exploration and a build-ready flow looked much the same in a Figma link. Engineers built from early ideas, stakeholders gave pixel feedback on concepts, and finished work was reopened late in the quarter.

I introduced five labels that designers attach to every design they share. Each one tells engineers and stakeholders how settled the work is and what kind of feedback helps. I aligned them with the quarterly Definition of Ready that product managers and I wrote, so a label shows where the work sits in planning as well as in design. Later, I added decision-making guidance that sets out which design decisions designers make alone and which need the Head of Design.

My role

As Head of Design, I wrote and own the labels and the decision guidance. I introduced the labels as a trial in early 2026 and the decision guidance in mid 2026, working with product managers, engineering leads and architects. The labels apply to design files, tickets and shared links.

The problem

Every design looked equally final, and the words we had made it worse.

“Final” or “signed off”
Sounds frozen and unquestionable, so people stop raising real risks.
“Mockup”
Sounds disposable, so careful decisions get overridden.
“Wireframe”
Is often ignored entirely, so early direction never gets challenged.

The labels

The labels describe intent and confidence, not visual polish or rank. Each one sets expectations on both sides: what the team can assume, and which feedback is useful now.

  1. Discovery

    Before design

    A playground of research and ideas, kept separate from design direction.

  2. Concept

    Exploring

    Several possible directions. Not ready to build, and major changes are expected.

  3. Proposed

    Converged

    A considered solution. Core decisions are intentional, and some constraints are still open.

  4. For development

    Source of truth

    What should be built. Open questions are called out explicitly.

  5. Production

    Refined

    Detailed design for high-impact work. Any departure from it should be deliberate.

Production design is not always needed. It is for features where the experience is critical to success, heavily used or heavily marketed. Designers re-label work as it matures, and add a short sentence when a label alone does not say enough, such as “Open to feasibility feedback”.

In practice

Each design or set of designs carries exactly one label, in three places, so the context travels with every link.

Design file cover
Proposed Certification renewal flow
File and page names

[PROPOSED] Certification renewal flow

[FOR DEVELOPMENT] Course status block

Ticket or chat message

Designer: Renewal flow is Proposed. Core behaviour is set; open to feasibility feedback on the bulk action.

These examples are illustrative, and the feature names are placeholders.

Aligned with the Definition of Ready

At the same time, I worked with product managers on an R&D quarterly Definition of Ready. It describes how product, design and engineering trios prepare work a quarter ahead, in four stages of growing confidence. I added the design view to each stage, so each label is a visible signal of how ready the work is.

Readiness stageConfidence the team needsDesign label
1. DefinedWe are solving the right problemDiscovery
2. ScopedWe are solving it in the right wayConcept
3. TestedThe direction reduces the biggest risksProposed
4. PlannedThe team can build it safelyFor development
After readyRefinement during deliveryProduction

Both aim to prevent the same problems: late feedback that causes major rework, unclear decision boundaries that send every choice for approval, and teams working around missing design direction.

Who decides

Once the labels made status clear, the next source of friction was who decides. Designers were unsure when they could act alone, and some decisions came to me for approval while others came too late. I wrote short guidance with three levels, tied to the same labels, with one principle throughout: default to autonomy and raise tension early.

  1. Level 1: Designer-led

    • Applying design system guidance
    • Refining layout, hierarchy and interactions
    • Trade-offs within agreed scope
    • Moving work through Discovery Concept Proposed

    Local, reversible and low risk, so go ahead.

  2. Level 2: Align with the Head of Design

    • Departing from the design system
    • Being unsure whether to keep legacy behaviour
    • Compromises that affect consistency or usability
    • Pushing for design debt fixes during delivery

    Discuss it to reduce friction later. This is not an approval step.

  3. Level 3: Shared decision

    • New patterns or interaction models
    • Precedent-setting work or accessibility risk
    • Unresolved tension between design, engineering and product
    • Major changes to work labelled Proposed For development Production

    Too big for one designer to own alone, so bring it in early.

Each request also names the kind of involvement it needs:

FYI
Visibility only.
Alignment
Discussion and guidance.
Decision required
A shared or escalated decision.

Naming the kind of involvement was as useful as the levels. Designers could bring work to me without it becoming an approval gate, and it was clear when a decision really was shared. Together the three tools answer different questions: the labels say how settled a design is, the Definition of Ready says how ready the work is, and the levels say who decides.

Outcome

  • Sign-off became easier. Stakeholders knew what they were being asked to approve, and when, so reviews converged faster.
  • Better feedback at each stage. Engineers raised feasibility on proposed work instead of discovering it during the build, and concepts stopped attracting pixel critique.
  • Shared language with product. The labels became part of the Definition of Ready, so product managers and designers use the same terms to describe how ready a piece of work is.
  • More trust with engineering. “For development” meant one thing, so teams could start work without second-guessing the design.
  • Less friction over decisions. Designers had clear room to act alone, and contentious calls reached me early and in the open, not as late escalations.

Reflection

The value comes from everyone using the same five words, so I would keep the set small and add a label only with the team’s agreement. Three things I would refine:

  • Separate confidence from stage. “Proposed” covers both early convergence and research-backed work. I drafted a confidence marker, directional or high confidence, to show validation without adding another stage.
  • Watch for levels becoming gates. Escalation guidance can turn into approval-seeking. I would check with designers that Level 2 still feels like a conversation.
  • Measure the rework. Tracking late-quarter design changes before and after would turn “easier” into a number.

Read next

Modernising an enterprise learning platform between two major releases