Enterprise Clinical Product
Schema Designer
A visual authoring product that helps medical writers create and edit clinical trial schema diagrams, capture structured metadata for each element, import prior schema designs, and generate draft study design text from the canvas.
Estimated 50%+ authoring time saved · Released 5 months ahead of schedule
Role
Product Strategy and Delivery
What I Owned
Strategy, roadmap, requirements, cross functional delivery, UAT, and release support
Outcome
Estimated 50%+ authoring time reduction based on user validation, with release five months ahead of schedule
THE FULL APPLICATION
One canvas, structured data, and generated text in a single workspace
Toolbar
Element palette, accelerators, and AI generation controls
Canvas
Drag-and-drop schema with auto-connect logic
Metadata Panel
Structured fields surfaced per element type
Generated Text
Draft study design text generated from the visual schema
Toolbar
Element palette, accelerators, and AI generation controls
Canvas
Drag-and-drop schema with auto-connect logic
Metadata Panel
Structured fields surfaced per element type
Generated Text
Draft study design text generated from the visual schema
This case study uses sanitized reconstructions of confidential enterprise work. Visuals are recreated for portfolio purposes under confidentiality constraints.
I translated medical writer needs into product requirements, prioritized capabilities based on user value and technical feasibility, aligned engineering and UI/UX through daily standups, led UAT with subject matter experts, and supported the product through release. I also ideated and created an import capability that transforms existing schema diagrams into editable, structured schemas within the product.
Product type
Enterprise clinical technology
Timeline
Two release cycles
Primary users
Medical writers
Looking for consumer product and recommendation work?
Explore SolaceThe Problem
Why schema authoring was difficult
Medical writers were recreating complex clinical trial schemas manually while managing related metadata and protocol content across disconnected tools.
The opportunity was not simply to make diagramming easier. It was to create a structured authoring system that could support the broader protocol workflow.
Before and After
From manual diagrams to a connected authoring system
Before
- Manual schema creation took multiple hours per study
- PowerPoint and Lucidchart with manual line-drawing
- Metadata captured in separate spreadsheets, disconnected from the diagram
- Existing schema diagrams could not be reused as structured, editable product content
After
- Cuts schema authoring time by at least 50%
- Purpose-built canvas with auto-connect logic that keeps diagrams clean automatically
- Structured metadata attached to each element, visible in a side panel
- Previous schema images can be analyzed and converted into editable Schema Designer elements
Strategy
How I defined the product direction
Roadmap and Prioritization
I translated user research and stakeholder feedback into sequenced priorities based on user value, technical feasibility, and cross product dependencies. I reconciled competing requests, made explicit scope and sequencing decisions, and adjusted or deferred capabilities when necessary to keep the product focused.
Constraints
Enterprise environment with clinical accuracy requirements, multiple stakeholders, and dependency on structured data shared across the broader Study Designer suite.
Target User
Medical writers responsible for creating and maintaining clinical trial schemas and protocol content.
Hypothesis
If medical writers could create schemas on a purpose-built canvas with reusable accelerators and connected metadata, authoring time would fall by at least half and the output would stay consistent across studies.
Central Pivot
The product shifted from a diagramming tool to a structured authoring system once it became clear that diagrams, metadata, and protocol text had to stay connected.
Reuse Existing Work
The schema import capability addressed a practical barrier to adoption by allowing medical writers to reuse previous work rather than manually recreating every element, metadata field, and footnote.
Research & Validation
How research shaped the product
Initial workflow discovery
5 medical writers and 2 clinical operations leads
Conducted early interviews with medical writers and clinical operations stakeholders to understand the existing schema creation process, identify pain points in manual diagram creation and disconnected metadata capture, and define the initial product direction.
Recurring prototype validation
Sessions throughout prototyping
Built a functional prototype and continued meeting with medical writers throughout prototyping. Used their responses to refine feature behavior, remove unnecessary metadata fields, expand accelerator options, and identify missing capabilities before development began.
Predevelopment stakeholder workshop
30+ stakeholders across 5 business groups
Led a workshop with more than 30 stakeholders across five business groups before development. Demonstrated the functional prototype, gathered feedback across differing workflows and priorities, and translated the findings into product changes, requirements, user stories, and acceptance criteria for engineering.
Continued feedback during development
UAT with subject matter experts
Throughout implementation I reviewed the product regularly and conducted UAT with subject matter experts before release, using their feedback to refine the canvas, metadata panel, accelerators, and the schema import workflow.
Supporting Evidence
40+ existing study schemas analyzed
I also analyzed more than 40 existing study schemas across multiple therapeutic areas and study phases to identify reusable structural patterns and common element types.
Product Decisions
Four product decisions, validated with real users
Turning complex trial logic into a flexible canvas
User Need
Medical writers spent hours manually drawing and connecting elements in PowerPoint and Lucidchart. Diagrams became unreadable as studies grew complex, and line management consumed more time than actual schema design.
Decision
Built a purpose-built canvas with drag-and-drop elements. The first version supported only manual line drawing between elements.
Tradeoff
Manual drawing was faster to build, but it carried the same line management burden writers already struggled with in their existing tools.
What Changed After Testing
After studying how applications like Lucidchart handle connections, we added the ability to connect a line automatically from four points on each element, keeping diagrams clean without tedious manual routing.
Capturing structured metadata without overwhelming
User Need
Schema elements needed structured data such as dosage, visit windows, and eligibility criteria, but this information lived separately in spreadsheets and documents, disconnected from the visual diagram.
Decision
Designed a curated metadata panel with only the fields that matched real workflows, removing fields that did not reflect how writers actually worked.
Tradeoff
Less flexibility for edge cases, but dramatically lower cognitive load and faster authoring for the majority of studies.
What Changed After Testing
Removed three metadata fields after SME review confirmed they were never used in practice. We also added a custom option to each element, so if any metadata was missed or writers wanted to capture and display something beyond the built-in fields, they had the flexibility to do so.
Designing reusable structures across studies
User Need
Writers repeatedly rebuilt the same dosing, randomization, and escalation patterns from scratch for every new study, duplicating effort and introducing inconsistency.
Decision
Designed accelerators that insert editable structures: starting points writers could modify, not locked templates that constrained the study design.
Tradeoff
Accelerators add toolbar complexity, but save significant time for common patterns and enforce structural consistency without rigidity.
What Changed After Testing
Initially prototyped with 2 accelerators, then added two more after SME testing and discussions surfaced additional common patterns. Released with 4 accelerator options.
Turning existing schema diagrams into structured, editable starting points
Featured decision
User Need
Medical writers already had schema diagrams from previous studies, but those schemas existed as static images or documents rather than editable, structured product data. Requiring users to rebuild every existing schema manually would preserve much of the work Schema Designer was intended to eliminate and create a barrier to adoption.
Decision
I ideated and created an import workflow that allows users to upload an image of a previous schema design. The product analyzes the image, identifies the schema elements, maps them to the element types defined within Schema Designer, extracts the associated metadata, and detects footnotes. Each element is assigned a confidence score that reflects how accurately the product extracted its information, and lower confidence elements are highlighted so users can direct their attention to them first. A side panel lists every detected element alongside its extracted metadata, so users can review and edit fields before placing it onto the canvas. Before the imported schema is placed onto the canvas, users can review and correct the result. They can change an element's type, edit its metadata, choose which metadata should be displayed, reorganize the element order, and review the detected footnotes. Once confirmed, the schema is placed onto the canvas as editable, structured content that users can continue building upon.
Tradeoff
A completely automatic import would have reduced the number of steps, but image interpretation can be imperfect and clinical schemas often contain study specific structures. Adding a review stage required more user involvement, but it preserved accuracy, transparency, and user control before the schema became part of the structured product workflow.
Product Value
This capability allows medical writers to reuse previous work rather than starting every schema from a blank canvas. It also supports the transition from static historical schemas into a structured authoring system while keeping users in control of how the imported content is interpreted.
Featured Capability
Importing an existing schema in five stages
Medical writers can reuse prior work instead of starting from a blank canvas, while staying in control of how the imported content is interpreted.
Upload
Upload an image of a previous clinical schema diagram.
Identify
The product identifies schema elements, associated metadata, and footnotes.
Confidence Score
Each extracted element receives an individual confidence percentage. Lower confidence elements are highlighted so users can focus there first.
Review and Edit
Users review, reclassify, edit fields, choose displayed metadata, and reorder the extracted content before anything is placed on the canvas.
Place on Canvas
The validated schema is placed on the canvas as editable, structured content for continued manual authoring.
Delivery
Cross Functional Delivery
I stayed involved from discovery through implementation and release. Functional prototypes gave medical writers, product stakeholders, engineering, and design a shared experience to evaluate before development. Detailed requirements, acceptance criteria, daily standups, product reviews, prototype validation, and UAT kept implementation connected to the validated product direction.
Alignment
Alignment
Functional prototypes let stakeholders click through the real experience, not interpret static screens. Working prototypes helped stakeholders reach decisions earlier by allowing everyone to evaluate the same product behavior.
Validation
Validation
I ran usability sessions on working prototypes. Medical writers could drag elements, edit metadata, and generate text, revealing real friction before development.
Handoff
Handoff
Instead of handing developers annotated mockups, I handed them a working reference. The functional prototype provided a working reference alongside product requirements and acceptance criteria, reducing ambiguity and unnecessary clarification during implementation.
Historical Schema Analysis
Analyzed 40+ existing study schemas across multiple therapeutic areas and study phases to identify reusable structural patterns and common element types.
SME Insight Summary
Conducted sessions with five medical writers and two clinical operations leads to map current pain points and validate the problem framing.
User Flow
Mapped the end-to-end authoring flow from protocol outline to schema creation, metadata entry, accelerator insertion, and text generation.
Prototype Iteration
Iterated through three prototype versions, each tested with SMEs. Each round refined the canvas behavior, metadata fields, and accelerator set.
Requirements & Acceptance Criteria
Translated the validated prototype into detailed user requirement specs, functional requirements, and acceptance criteria for the development team.
UAT Finding & Resulting Change
During UAT, writers expected generated text to respect element ordering in the diagram. Updated the generation logic to match before release.
Predevelopment Stakeholder Workshop
Led a workshop with more than 30 stakeholders across five business groups before development, demonstrating the functional prototype and translating feedback into requirements, user stories, and acceptance criteria.
Cross Product Consistency
Collaborated with the team members building the other products within the Study Designer suite to align patterns, components, and interactions, so users moving through the overall Study Designer product felt one cohesive application rather than a collection of separate products.
Delivery Impact
How functional prototypes changed the handoff
Old Handoff Process
- 01Annotated mockups handed to developers
- 02Developers interpret static screens
- 03Questions and clarification cycles
- 04Revisions based on misunderstandings
- 05Weeks of back-and-forth before build starts
Functional-Prototype Process
- 01Functional prototype with real interactions
- 02Developers click through actual behavior
- 03Requirements embedded in the prototype
- 04Less clarification needed with a working reference
- 05Build proceeds with a working reference alongside requirements
The functional prototype served as both a working reference and a validation. Developers built against a working reference, not an interpretation of static screens, reducing ambiguity and unnecessary clarification during implementation.
Product Walkthrough
The product in motion
Building a trial schema
Medical writers drag, connect, and arrange study elements visually. Auto-connect logic keeps diagrams clean without manual line-drawing.
Applying reusable accelerators
Accelerators insert editable dosing and randomization structures, so writers stop rebuilding the same patterns from scratch.
Generating draft study design text
Generated text turns the visual schema into draft study design language.
Generated Study Design Text
Generated from schema
Eligible participants will be randomized in a 1:1 ratio to receive either the investigational agent or placebo. Randomization will be stratified by biomarker status and prior therapy line. Treatment continues until disease progression or unacceptable toxicity.
Importing and extending an existing schema
Users upload an image of a previous schema and the product converts it into editable Schema Designer elements. It identifies element types, associated metadata, and footnotes, and assigns a confidence score to each element showing how accurately the information was extracted. Lower confidence elements are highlighted so users can direct their attention there first. A side panel lists every detected element with its extracted metadata, so users can review and correct classifications, edit fields, control what is displayed, and reorganize the structure before the confirmed schema is placed onto the canvas, where they can continue editing and adding to it.
Results and Learnings
Est. 50%+
Authoring time reduction based on user validation
Estimated, based on user validation
5 months
Ahead of schedule
2 cycles
Release cycles with UAT driven changes
Estimated 50%+ authoring time reduction based on user validation, released five months ahead of schedule.
Schema Designer gave medical writers a purpose-built canvas, reusable accelerators, and structured metadata, while establishing a foundation for connected tools across the broader Study Designer suite.
What I Would Measure Next
- 01Authoring time per schema, before and after Schema Designer, across the full medical writer population.
- 02Reuse rate of saved schemas and accelerators across new studies.
- 03Adoption and consistency across the broader Study Designer suite.
- 04Time from validated concept to released capability.
Next Project