Home care · B2B SaaS

UX design for complex home care operations

Turning dense operational data, business rules and interdependent workflows into clear, implementation-ready interfaces.

Minimal system-style diagram on a muted blue background. Ally workflows and EVV workflows are positioned around a central label reading structured operational interfaces.
Two operational workflow areas shaped one design challenge: turning complex home-care processes into structured interfaces.

Project context

Client and product
Ally MS · Ally platform · US HealthTech B2B SaaS for home care operations, electronic visit verification, billing and insurance claims.
My role
UX Designer responsible for client-facing discovery, system mapping, workflow design, low-fidelity UX and delivery coordination.
Team and ownership
Two UX Designers shared discovery, workflows and low-fidelity design. A UI Designer owned the UI Kit and final visual design.
Timeframe
May–November 2023.
Engagement
System mapping, UX modernisation and workflow design aligned with the development roadmap.
Core constraint
Required operational data and business rules could not simply be removed or changed.
  • System Mapping
  • Information Architecture
  • Complex Workflow Design
  • Data-Heavy Interfaces

Designing within a larger operational system

Ally is an operations and billing platform built for US home-care registries working with independent caregivers. It supports client and caregiver records, scheduling, visit verification, payment processing, caregiver payments, insurance claims and reporting.

Selected Ally and EVV workflows needed to be mapped and modernised ahead of scheduled development sprints. We clarified dependencies with domain experts and translated them into clearer information hierarchies, interactions, system states and low-fidelity UX flows.

Challenge

Extensive operational data, strict rules and linked records had to remain understandable without losing necessary context.

Validation

Designs were reviewed with product stakeholders, administrators, superadmins, engineers and sales specialists.

Delivery

Completed UX flows were handed to the UI Designer and prepared for implementation against the development roadmap.

The section of the system we explored and redesigned

Selected workflows within a much larger operational platform were mapped for redesign, with adjacent functionality reviewed only as needed to understand data, rules, states and downstream dependencies.

Scope map divided into Ally and EVV. Ally shows a smaller UX scope for claims scheduling and generation, Accounts Receivable, caregiver payments and messaging, alongside adjacent product navigation. EVV shows a broader observed scope across operational and administrative areas. Both connect to related operational areas, while system-to-system data flow remains outside the project scope.
Project scope map distinguishing workflows mapped or redesigned from adjacent functionality reviewed for context. Ally represented a smaller project slice; EVV covered a broader administrative scope. This reflects the available project scope, not the architecture of the entire system.

System complexity across records, rules and lifecycle states

Each interface exposed a larger system of records, rules, states and downstream consequences. The operational logic and required data had to remain intact, while the interface had to make dependencies, validation and next actions understandable.

Before simplifying the Visit view, we decomposed the existing interface and documented the relationships and open questions that shaped the redesign.

An anonymised mapping board for an EVV Visit. Readable section headings organise blurred interface fragments, and arrows connect grouped comparisons to related examples.
Before redesigning the Visit workflow, we decomposed the existing interface into its major sections and mapped how the surrounding elements related to one another. This exposed the questions that required client validation.

Reducing information density without losing operational meaning

The Visit details view had to combine operational states, related records, rejection context and recovery actions. Early low-fidelity versions included the full diagnostic layer within the same expanded record.

The challenge was to decide what needed to remain visible, what could move into linked detail and how to preserve the context required for action.

An anonymised low-fidelity EVV Visit screen with one expanded record containing several blurred information groups and controls beneath a summary.
An expanded low-fidelity Visit view helped the team assess the full information burden before deciding what needed to remain visible. All expandable sections were intentionally opened for this comparison.
An anonymised final EVV Visit screen showing one expanded record, a concise summary, a separate lower section and a small set of visible status indicators and actions.
The delivered interface reduced the amount of information shown at once while preserving the context and actions administrators needed to review a Visit.
Complete diagnostic layerRelated states, explanations and actions shown together.
Implementation-ready hierarchyA prioritised summary with links to relevant detail.
Primary viewSupporting information moved behind links or removed from the main task.
AttentionCritical context and next actions prioritised through grouping and progressive disclosure.

The two UX Designers explored grouping, progressive disclosure and action hierarchy. The Product Manager made the final decision about which system information and functions could be omitted. The UI Designer translated the agreed low-fidelity direction into the final visual interface.

Multiple processing states

A Visit could surface several independent processing states at the same time. They could not be collapsed into one universal status without losing the meaning of where an issue occurred and what could happen next.

An anonymised final EVV Visit screen with six numbered annotations marking status indicators across one expanded record.
One Visit context could surface several independent processing states at the same time. The annotations show why the interface needed to keep them distinct without reducing the workflow to a single status.
  1. 1–2 Primary row states.
  2. 3–5 Related states inside the expanded summary.
  3. 6 Downstream processing state.

Validation changed with the workflow stage

Profiles could be created with the minimum information required to begin and completed later. Rules evaluated additional requirements when the data became necessary in the workflow.

  • Requirements could apply at different intervention points rather than at account creation alone.
  • Missing or invalid information could trigger a warning, prevent an operation or create an exception, depending on configuration.
  • Administrators needed enough context to identify the affected record and continue or resolve the process.

Designing a multi-level Accounts Receivable workspace

Accounts Receivable brought several levels of financial information into one operational workspace: account-level totals, balances grouped by care recipient and payer, invoices, service details, transaction history and payment allocation.

Our task was not to redefine the financial rules. It was to let administrators move from an overview to a specific invoice or transaction, understand the scope of an action and keep the surrounding account context visible.

One workspace, two aggregation perspectives

The dashboard supported two ways of finding the same receivables: through care recipients and through payers.

Anonymised Accounts Receivable dashboard with summary values and two related data views, including one expanded row showing an additional level of information.
The final dashboard brought two complementary views of receivables into one workspace while preserving the relationship between summary and detailed information.

Keeping transaction history in context

An invoice connected account-level data with service details and a history of payments, adjustments and other billing events.

The UX and final UI retained the same interaction model: select an invoice, expand it in place, then switch between Service details and Transaction history. This kept the parent invoice and surrounding records visible while exposing deeper information only when needed.

Side-by-side anonymised comparison of a UX wireframe and final UI. Both show a list with one expanded row and a nested transaction-history view.
The UX structure and final UI use the same expandable model, keeping transaction history in context without moving the user away from the invoice list.

Making payment allocation visible

After selecting invoices, an administrator could assign an amount to each one while keeping the total cheque amount and amount remaining visible. Additional exception handling appeared only where an individual invoice required it.

Anonymised final Post payment interface with three invoice sections, per-invoice payment fields, one additional reason control, a cheque total and the amount remaining.
The final interface supported allocating one payment across selected invoices while keeping the total and remaining amount visible.

Supporting pattern

Turning parameters into a guided setup

An early claims-automation concept presented cadence, frequency and execution times as a parameter-led form. We reorganised the setup around three questions: What time? How often? And, for a weekly cadence, What days of the week?

Conditional controls appeared only when relevant, while a readable summary made the schedule explicit before saving. A compact status message then kept the active cadence and last completed run visible.

Early claims-automation form organised around time period, frequency and execution times.
Final guided claims-automation setup asking when and how often automation should run and revealing weekday choices for a weekly cadence.Status message showing the active claims-automation schedule.

The configuration evolved from a parameter-led form into a guided setup with conditional inputs, a readable schedule summary and persistent status feedback.

Reflection

This project reinforced that design practice has to adapt to the complexity of the product. In a system shaped by business rules, integrations and backend constraints, progress depended on learning from people who understood the system deeply, then translating that knowledge into a usable interface model.

My role was to turn specialist knowledge into clear information hierarchy: distinguish actionable from informational data, editable from read-only content, expose the right states and links, and reduce information density without hiding the operational context required for decisions.

Evidence boundary

The case study covers selected workflows rather than the complete platform. Without direct access to commercial end users or verified outcome metrics, the evidence is limited to expert and stakeholder validation, approved UX direction and final delivery.