Case Study — Healthcare / Enterprise Design System

CHG Healthcare — One Schema, Two Applications

CHG Healthcare runs two provider credentialing applications — one for physicians placed through its CompHealth and Weatherby Healthcare staffing brands, another for allied health providers — being unified into a single schema-driven system, built and configured through Modio, the editor CHG's credentialing team uses to define its forms.

The two applications shared almost no common data model, and neither matched how fields would need to be structured once configured inside Modio, the editor CHG's team builds its forms in. Before a single wireframe could be drawn, twenty sections of every application needed to be compared field-by-field, and twenty-four architecture, product, and legal decisions needed an owner and a recommendation. This case study is about that decision-making — not the screens, which are still in progress.
Role
Senior UX Designer
Focus
UX Audit, Design System Architecture & Decision Framework
Company
CHG Healthcare
Collaborators
Product, Engineering, Legal, Credentialing Operations
Status
In progress — Phase 1 decisions resolved
Tools
Figma, FigJam
2
Applications unified onto one schema
20
Sections audited, from Personal Info to Final Submit
24
Architecture, product & legal decisions documented with an owner
2-tier
Configurability model: locked, configurable

Two applications, no shared schema

Each application was built separately, at a different time, for a different audience. Neither shared an underlying data structure with the other — which meant every future improvement would have to be built twice, or the team would first have to work out how one schema could serve both, configured inside Modio, the editor the team builds its forms in.

CHG Physician (Locums)

An 18-page credentialing application used by CHG's staffing brands, CompHealth and Weatherby Healthcare, to place physicians on locum tenens assignments.

CHG Allied

A 16-page application for allied health providers — nurse practitioners, physician assistants, registered nurses, and other non-physician roles — with its own section structure.

Compare every field before designing a single screen

The first deliverable wasn't a wireframe — it was a complete field-level audit of all 20 sections across the two applications, benchmarked against how each field would behave once configured inside Modio, the editor. Every row was classified with a badge system so engineering and product could see, at a glance, what was shared, what was unique to each application, and what still had no decision.

Field-by-field comparison 20 sections audited 5-badge classification system Cross-checked against discovery research
Field Badge CHG Physician In Modio (the Editor)
Primary Specialty Decision needed Dynamically controls the Clinical Capabilities page No equivalent — the editor isn't specialty-aware
NPI Number Differs Open text The editor validates in real time against the NPPES registry
Smart entity lookup Decision needed Not present The editor offers predictive lookup from NPPES, ACGME & AHA hospital directories
CAQH section Native to the editor Not present Provider ID, attestation dates, account status

Three findings that changed the scope

Comparing field-by-field, instead of designing directly on top of assumptions, surfaced problems a surface-level redesign would have missed.

"Allied" isn't one form

The Allied application actually covers at least two distinct provider groups — Cota/EEG Tech/Physical Therapy versus Nurse Practitioner/Physician Assistant/Psychology — with different depth in Education, attestations, and professional liability. That wasn't visible until the fields were compared one by one.

Specialty drives everything downstream

Selecting the Primary Specialty dynamically controls which clinical questions appear later. It's the most foundational decision in the whole project — and it can't be added later without a schema rebuild.

A legal contract blocked data enrichment

Legal flagged that an external data provider's contract prohibits sharing data directly with third parties — meaning auto-enrichment inside the editor needed legal review before it was built, not after.

One parent schema, two applications, one editor

The answer to "how does one schema serve two different applications, built through the same editor" was a two-tier configurability model. Every form field is classified as either locked or configurable, and that classification — not the individual screen — is the real design system behind the platform.

Tier 1 — Locked

Compliance-driven

Cannot be removed by either child template. NPI, DEA, malpractice questions, attestation and signature. Renders identically no matter which application shows it.

Tier 2 — Configurable

Toggleable per template

Fields each application can toggle, relabel, or adjust — Medical Training, the Life Support certification list, Professional Organizations — depending on how they're configured inside the editor.

Modio, the team's editor, is where the shared schema gets built. CHG Physician and CHG Allied each render it as their own application.

24 decisions, each with an owner and a recommendation

Every decision was framed as a clear choice with options, each one's impact, and a team recommendation — organized from the most architecturally critical (blocks schema design) to the most tactical (copy and microcontent).

7
Architecture — critical, blocks schema design
7
Product — needs Ops, Legal or Finance input
3
Page structure — UX Design decides
5
UX & copy — UX Design resolves directly
2
Legal & compliance
DECISION 1A · CRITICAL

Does specialty route content, or is it just data?

Chosen: routing. Higher Phase 1 complexity, but delivers CHG's core value proposition and can't be added later without a schema rebuild.

DECISION 1F · CRITICAL

Should the application fill itself in?

Chosen: full smart lookup. Predictive lookup of schools, hospitals, and employers against NPPES, ACGME, and AHA directories, cutting redundant provider data entry.

DECISION 1G · CRITICAL

Can the schema, once configured in the editor, use the same enrichment data?

Chosen: legal review in parallel. A third-party data contract limited how enrichment could be used once configured inside the editor. Rather than block Phase 1, legal review runs in parallel while the rest of the schema keeps moving.

DECISION 2B · HIGH

How many variants does "Allied" actually have?

Chosen: sub-templates by provider type. An audit finding, not an original requirement — the Allied child template branches by provider type instead of using one form for everyone.

None of this could be simplified for convenience

This wasn't a consumer product flow where a confusing field can just be cut. It was a regulatory application shared across product, engineering, legal, and credentialing operations teams, each with their own reason not to budge.

Compliance isn't negotiable

Tier 1 fields — NPI, DEA, malpractice questions, attestation — are mandatory no matter which child template renders them. No design decision could touch them.

Building next to a moving target

A CV auto-extraction pipeline was already in active engineering development in parallel. The new platform had to plug into that pipeline's event schema, rather than design in isolation.

Four teams, one timeline

Product, engineering, legal, and credentialing operations all had to agree before wireframe design could start — and architecture decisions had to be resolved first because they're the most expensive to reverse once development begins.

24 decisions resolved into an engineering-ready schema.

The audit and decision brief became the shared reference between product, engineering, legal, and credentialing operations — unblocking wireframe and schema design while keeping compliance-critical fields untouched across every child template.

Unblocked
Schema design could begin once architecture decisions were resolved
Reusable
The locked/configurable split isn't tied to one application — any future form built in the same editor can reuse it
Traceable
Every decision keeps its rationale and owner, not just the outcome
Most of this case study is decisions, not screens. That's the nature of platform-architecture work — the highest-leverage design choices here were made months before a single wireframe existed, and getting them wrong would have meant rebuilding the schema after engineering had already built on top of it.
A note on visuals: the architecture diagram on this page is an original illustration I made to explain the parent/child model — not a Figma export. CHG Healthcare's Phase 1 visual screens are still in progress and will be added here as they're finalized. Internal contributor names have been generalized to role.
All work
All Work