Restoring Trust at Scale: A Compliance Redesign for 10,000+ Developer Organizations

Client
Confidential (Developer Tools Domain)
Timeline
6-month iterative process
Role
UX Designer · State Modeling · System Architecture · Research · User Testing · UX Writing · Accessibility · Engineering + PM Partnership
Tools
Figma, Figjam

I designed a systems-level compliance experience that helped 70% of extensions reach certification across 10,000+ organizations within 90 days.

‍

‍

📖 Overview

Marketplace compliance had become difficult to interpret at scale. Signals were inconsistent, rules were scattered across disconnected tools, and certification status was not clearly visible at the extension level, where developers and admins needed it most.

I redesigned the compliance experience around a clearer trust model: distinct publisher and extension signals, plain-language states, automated routing, risk-based prioritization, accessible indicators, and guidance that worked across dense marketplace surfaces.

Goal Create a unified, predictable, trustworthy compliance system.
Outcome 70% certification adoption across 10,000+ organizations within 90 days.

‍

🧩 The Problem

Compliance had become brittle and inconsistent, creating systemic failures across publisher verification, extension certification, admin triage, and reviewer workflows.

“With hundreds of extensions, everything looks risky. I can’t tell what to fix first.”
— Marketplace Admin
Core insight Users did not need more governance. They needed clearer language, predictable states, and a shared mental model for trust.

‍

🧑‍🤝‍🧑 Team & Collaboration

I partnered across PM, engineering, accessibility, legal, and compliance to ship a scalable, standards-aligned system.

‍

🔍 Research & Insights

I interviewed marketplace admins, engineers, PMs, and compliance reviewers to understand why trust decisions felt inconsistent, slow, and difficult to explain. Four insights shaped the redesign.

Design Principle Trust at scale comes from shared language, not added complexity.
Legacy dense extension list
Before

Flat, dense list: every extension looked equally risky.

Prioritized extension list
After

Prioritized, trust-first design: high-risk items rise to the top.

‍

Compliance State Model

1. Four-State Compliance Model

I replaced legacy risk language with four plain-language states that made compliance status easier to recognize, compare, and act on across dense marketplace surfaces.

Compliant

Meets all requirements.

Action Required

Next steps are shown inline.

Non-compliant

High risk. May be disabled until resolved.

Disabled

Temporarily turned off as a safety fallback.

‍

Layered Trust Signals

2. Publisher & Extension Trust Signals

I designed a layered trust language that separated publisher identity from extension safety while still allowing both signals to read as one cohesive system across the marketplace.

Publisher Trust

Indicates whether a publisher meets verified identity and domain requirements.

Publisher identity uses a verification badge (✔︎), while extension trust uses shield-based compliance states. These signals work together but represent different layers of trust.

Extension Trust

Shows whether an extension has passed signing, permission, and update-safety checks.

Marketplace Example
Marketplace cards showing CORP-certified publishers with a verification badge and trust shield.
A real surface where both layers of trust appear together: publisher verification (✔︎) and extension compliance shields.

Key decisions

  • Replaced color-only indicators with evergreen icons and labels.
  • Optimized icon readability at small sizes for dense surfaces.
  • Applied consistent visuals across all Marketplace contexts.
  • Separated publisher and extension safety signals.
“The shield makes it obvious. I know exactly what is safe.” Admin Tester

‍

Risk-Based Prioritization

3. Prioritized triage

I introduced severity-based ordering so high-risk items rose to the top instead of getting buried in dense extension lists. Priority was communicated through labels, iconography, and supporting text so the system did not rely on color alone.

‍

Verification & Routing Logic

4. Automated Verification & Routing

The system identifies publisher type automatically and routes users into the correct path. This made naming rules, domain verification, certification options, and trust signals more predictable across 1P and 3P publisher flows.

1P Publishers
  • Naming rules: Strict internal-affiliation conventions
  • Domain verification: Corporate domain required
  • Identity signal: Verified publisher badge (✔︎) once CORP identity is confirmed
  • Certification path: Accelerated
3P Publishers
  • Naming rules: Avoid internal-style prefixes
  • Domain verification: Optional but guided
  • Identity signal: No verification badge; trust is instead communicated at the extension level through compliance shields
  • Certification path: Transparent with opt-out
Automated routing flowchart

‍

Shift-Left Quality Checks

5. Earlier Compliance Guidance

Compliance checks previously surfaced late in the publishing flow, creating rework, blocked approvals, and back-and-forth with reviewers. I shifted guidance earlier in the lifecycle so publishers could catch issues during authoring and pre-submission instead of discovering them after upload.

Before: Late-stage checks
Authoring
No guidance; policies hidden in docs.
Upload & Review
Issues = rework loops
Violations caught after upload, blocking approvals.
Rework
Back-and-forth with reviewers; unclear blockers.
Approval
Slow, unpredictable path to compliant state.
After: Shift-left quality
Authoring
Inline hints
Real-time guidance while editing manifests and metadata.
Pre-submission checks
Critical checks
Blockers vs warnings separated with clear, actionable errors.
Upload
Clean, policy-aligned packages hit the pipeline.
Approval
Faster, predictable approvals with fewer review cycles.

Moving checks earlier reduced late-stage rework, cut down back-and-forth, and made approvals more predictable for both publishers and reviewers.

‍

🧪 User Testing Strategy

I tested two high-friction compliance flows to evaluate comprehension, confidence, and decision-making at critical points. Sessions used scenario tasks, live observation, think-aloud feedback, and post-task reflection to identify where the model broke down.

Flow 1: Initial Setup

Creating a publisher → uploading an extension → assigning roles.

Scenario Tasks Live Observation Post-Task Reflection

Flow 2: Reclassification

Resolving issues or opting out of certification, with downstream impact checks.

Think-Aloud Critical Incidents Confidence Ratings
1
Observation
Level confusion

Publisher vs extension certification blurred.

Outcome
Clear levels

Copy separates publisher vs extension certification.

2
Observation
State uncertainty

“Action Required” and “Non-Compliant” lacked clear outcomes.

Outcome
Outcome ladders

States now show consequences and next steps.

3
Observation
Icon ambiguity

Shield meaning unclear without labels.

Outcome
Labeled trust

Shield is paired with evergreen labels and helper text.

4
Observation
Routing surprise

1P vs 3P paths felt unexpected.

Outcome
Transparent routing

Inline explanations for 1P and 3P paths.

5
Observation
Triage overload

100+ flagged extensions with no prioritization.

Outcome
Prioritized triage

High-risk items surface first through severity-based ordering.

Clearer language, visible routing, labeled trust signals, and risk-first ordering reduced cognitive load and helped users make faster, more accurate decisions.

‍

🔄 Iteration & Validation

Early versions used similar trust signals for both publishers and extensions, which blurred the distinction between verified identity and extension-level compliance. High-urgency extensions also disappeared inside dense publisher lists. Through multiple rounds of testing, I separated the signal types and restructured the UI around how admins actually scan for risk.

Before: Publisher and extension signals were mixed, using the same shield, and urgent extensions disappeared inside dense tables.
Before design with shared shields and dense tables
After: Verified publisher badge (✔︎) for identity, shield-only states for extensions, labeled state cards, and role-guided resolution paths that surface the highest-urgency items.
After design with distinct badges and guided paths
Admin
“I can finally tell what needs action. It used to take 20+ minutes just to scan.”
PM
“Separating publisher badges from extension shields made the model click instantly and cut our documentation overhead.”

‍

📊 Outcomes

Before

Confusing signals
No prioritization
Unclear publisher types
Fragmented tooling

After

Unified compliance state model
40–50% faster triage
Automatic 1P/3P routing
Cohesive shields + labels
70% certification adoption in 90 days

‍

🧠 What I Learned

Principle

Governance ≠ bureaucracy

Clarity beats control.

Pattern

State models need hierarchy

Clear states → faster cognition → fewer errors.

System UX

System UX is language

Consistent words create consistent decisions.

Collaboration

Partnership wins

Embedding with engineering ensured scalability, not just polish.

Accessibility

Accessibility is trust

Evergreen icons outperformed color-only cues for every user.

‍

🏁 Reflection

This project changed how I approach complex systems. I learned that trust at scale does not come from adding more controls; it comes from making decisions easier to understand, explain, and repeat. By designing a shared mental model instead of just another interface, we helped thousands of organizations move through compliance with more confidence.

‍

My inbox is open.

Thanks for dropping a line! I'll be in touch shortly.
Oops! Something went wrong.

© 2026 Kendra Howe