IT: +39 338 62 46 214
BioXpert — Pharma Development Software
introduction
BioXpert is an internal web platform built for a global pharma company to unify and govern all Bio-X initiatives and scientific studies to build new drugs. Before BioXpert, data and project plans were scattered across multiple tools, making decision-making slow, opaque, and error-prone.
With BioXpert, we centralized planning, prioritization and resource allocation into a single tool — giving teams a shared view over feasibility, impact, risk and progress.
The result: faster, data-driven decisions, improved cross-team collaboration, and full transparency across Bio operations.
Client
Date
Categories
Tags
1
% faster decisions
1
% more data consistency
As a Lead Designer within a large and structured organization, I partnered with product owners, technical leads, business analysts, and research teams to transform fragmented workflows into a unified platform. The goal was to establish clear decision-making standards, streamline scientific planning, and connect teams through consistent, validated flows. Along the way, I created a dedicated design system aligned with the corporate style and iterated prototypes with users, improving clarity, reducing duplication, and ensuring the tool could scale reliably within the broader digital ecosystem.
About the project
In pharmaceutical research, decisions about which experiments to run and which activities to prioritise are costly and long-horizon. At this company, the teams making those decisions had no shared system — every project started with a blank spreadsheet, rebuilt from scratch, with no common structure or ownership model.
BioXpert was built to fix that: a platform for BioX teams to plan and prioritise research activities with shared structure and a single source of truth. I joined as sole UX designer for a 4-month end-to-end engagement.
— Timeline: 4 months · fast-paced delivery
— Role: Sole UX Designer · Lead Designer
— Scope: Research, UX/UI design, art direction, design system
My responsibilities
Design team
— Andrea Mabellini — Sole UX Designer · Lead Designer · end-to-end from research to handoff
Key stakeholders — BioX Scientists · Product Owners · Operations Leads · Data Managers · Tech Leads
Business trigger — BioX had no dedicated software for activity planning. Research decisions ran on spreadsheets with no version control and no shared ownership model. The inconsistency was creating downstream data quality failures that delayed study timelines and regulatory validation cycles.
Problem statement
Before BioXpert, teams depended on fragmented spreadsheets and isolated tools to plan and prioritise BioX activities.
The strategic problem
— No shared data structure — every team used its own naming conventions and formats; no common definition of what an “activity” was
— No version control — which spreadsheet was current? No one knew
— No ownership model — data passed between roles with no formal handoff or accountability
— Decision quality degraded — strategy calls made on incomplete or outdated data, with no audit trail
Example use cases
BioXpert supports diverse scenarios across BioX teams — from evaluating feasibility to comparing strategies or assessing resource impact.
— Developing a new BioX strategy — historical view of deprioritised activities, risk capture, and prioritisation tool for project teams and budget requests
— During study conduct — extract biomarker strategy, support financial controllers with BioX expenses, prioritise work requests
— Single source of truth — eliminates the need for multiple tools and information silos across all functions working on a study
Workflows
The platform is built around three primary workflows: activity creation, feasibility review, and prioritisation — each structured around a decision moment, not a tool feature.
Key finding: the most critical workflow wasn’t creation — it was comparison. Scientists needed to see multiple activities side by side before prioritising. The dashboard emerged directly from this finding as the primary entry point, not a secondary reporting view.
Personas
5 BioX personas defined from 8 interviews and workshops: scientists, activity owners, feasibility leads, financial controllers, and strategic leads.
Key finding: roles had conflicting needs. Scientists wanted flexibility and speed; controllers needed auditability and fixed structures. This tension drove the role-based visibility model — one platform, different field sets and permissions per role — rather than a compromise that would have satisfied no one.
Information architecture
BioXpert’s IA mirrors the natural hierarchy: Project → Study → Activity → Sub-activity.
Key finding: early wireframe testing showed users understood this structure immediately — it matched their existing mental model. The previous spreadsheet approach had failed precisely because it imposed a flat structure on a hierarchically organised problem. This validation accelerated stakeholder sign-off on the IA without additional rounds.
Workshop
I ran a series of collaborative workshops with scientists, project managers, and technical leads to align teams and surface the real challenges behind BioX workflows.
— Persona definition — built shared understanding of who uses the system and why
— Journey mapping — traced where data was lost, duplicated, or misinterpreted
— Pain-point prioritisation — ranked problems by frequency and impact
— Rapid ideation — validated early directions before any screens were designed
Workshops grounded every design decision in cross-functional insight — not assumptions.
Key learnings After Research
Research confirmed the structural gap: not a usability problem, but the absence of a shared data model across the BioX lifecycle.
A key pivot
By interview five, a scientist described rebuilding the same activity structure from a blank spreadsheet every new study — because nothing carried over and naming conventions weren’t shared. The problem wasn’t the UI. It was that no one had defined what an “activity” was. Brief shifted: from “design a planning tool” to “define the data structure that makes planning possible.”
Research: Hypothesis vs Reality
— Initial assumption — scientists needed a better interface for existing planning tools
— What research revealed — users rebuilt context from scratch at every study stage; no shared ownership, no version control
— Shift in scope — from “redesign the tool” to “define the data governance layer”
Key design decisions
— Hierarchical IA — project → study → activity → sub-activity mirrors how BioX teams think. Rejected flat tag-based structure: caused the same duplication problem users already had in spreadsheets.
— Standardised activity creation — single structured entry replacing emails and shared docs. Rejected flexible open forms: too inconsistent for downstream validation.
— Dashboard as primary entry — consolidated comparison view, not a deep-drill start. Rejected tool-by-tool navigation: replicated the fragmentation users were escaping.
Wireframes
Low-fidelity wireframes validated IA and core flows before any visual design — mapping the activity hierarchy, entry points, and role-based views.
Key finding: the first wireframe showed all activity fields on one screen. Feedback in the initial review was immediate: “too much, too soon.” Progressive disclosure became the structural principle — each view shows only what’s needed for the current decision step.
Colours
Colour palette built for a data-heavy scientific environment: neutral backgrounds for readability, structured use of accent colour to signal action states and priority levels without introducing noise in dense comparison views.
Typography
Type system optimised for data tables and nested hierarchies — a common pattern in BioX workflows. Clear size steps between labels, values, and section headers reduce the cognitive effort of reading across complex grids.
Spacing & Grids
8px grid applied consistently across forms, tables, and modals — the three primary interface patterns in BioXpert. Spacing decisions were driven by readability in dense activity-entry and comparison contexts.
Inputs
Form components designed for structured data entry across study, activity, and sub-activity levels — mandatory field logic, validation states, and role-based field visibility all specified at component level.
Tabs
Tab pattern used to segment the activity view by stage (Feasibility, Impact, Prioritisation) — keeping each evaluation step visible and navigable without losing the overall activity context.
Activity creation
The activity creation flow was the highest-stakes design challenge — the single entry point replacing emails, spreadsheets, and verbal conventions across teams.
Key finding: the first version had 18 fields on one screen. Testing showed users stalled at step 2. Reduced to 6 required fields at creation; optional fields surface at feasibility and validation stages. Completion rate in testing went from ~40% to ~90%.
Dashboard
The dashboard is the primary entry point — a consolidated comparison view of BioX activities across feasibility, impact, and prioritisation status.
Key finding: before BioXpert, prioritisation decisions were made in meetings, from memory, with no visual comparison surface. The dashboard didn’t just improve a UI — it replaced a process that had no digital equivalent at all.
Results
BioXpert delivered measurable improvements validated through two rounds of moderated usability testing with 6 BioX scientists and data managers, comparing task performance on the existing system against interactive prototypes of the redesigned flows.
— +36% faster task completion on key decision workflows
Measured via timed task sessions: participants completed approval and prioritisation flows in an average of 4.1 min on the new prototype vs 6.4 min on the existing system
— +40% reduction in data input errors
Assessed through a data quality audit of submitted entries during testing: standardised forms and mandatory field validation reduced inconsistencies compared to free-form entry in the legacy system
— −25% duplicated activity entries
Identified through a structured review of test submissions: the single-entry creation flow with deduplication prompts eliminated the parallel-entry pattern observed in the as-is workflow
— −30% navigation steps on critical tasks
Measured by counting interaction steps across task scenarios: the redesigned IA reduced average steps from 11 to 7.7 on the three highest-frequency user journeys
Methodology: moderated task-based usability testing (n=6), think-aloud protocol, pre/post task comparison against legacy system flows, error logging and step-count analysis.
Use any form
Design impactful user experiences with ease—everything you need is at your fingertips.
Feel free to reach out for a project inquiry, a coffee chat, or a consultation. Let’s explore how I can help you achieve your design goals. Whether it’s crafting impactful experiences or discussing innovative ideas, I’m here to collaborate and bring value to your vision.