Background

Ozette’s platform, originally designed to harness AI-driven analysis for unlocking the immune system, initially served as a proprietary tool for our data scientists to analyze clinical data on behalf of partners. However, as Ozette’s partnerships grew, so did the demands on data processing and result generation, challenging the platform’s efficiency.
‍

The Problem

While Ozette’s data scientist-led model worked initially, scaling brought challenges. The growing volume of data and contracts caused delays in result generation, increasing the workload on our data science team and creating operational bottlenecks.

Additionally, partners lacked visibility into the data analysis process, limiting their control and preventing efficient collaboration.This slowed partners’ access to their results.
‍

The Solution

SOLUTION DEATILS

BACKGROUND

Our Proprietary Software Was Slowing Us Down

Because our platform was proprietary, clinical partners had limited interaction with it. They provided their data, and we managed the entire data analysis process.

Their only involvement with the platform was during scheduled video calls, where data scientists would seek feedback or approval before moving forward.

But this model caused problems for both users.

From these pain points, we saw an opportunity to explore how to incorporate collaboration into the data analysis workflow.

PRELIMINARY RESEARCH

So, what can collaboration look like?

We studied how similar companies balanced internal ownership and partner collaboration in regards to data access, feedback, and communication.

These findings led to a reframe of our design question to:

“How can we support partner collaboration on the platform while protecting our intellectual property?”

USER RESEARCH/SYNTHESIS

Merging Workflows in A Multi-User System

Now that we we had a sense of direction;  we needed to first understand the workflow for both user groups and identify opportunities for collaboration.

I led the research efforts for understanding internal workflows and supported the PM in conducting walkthroughs and interviews with external partners.

Based on the findings, I mapped the end-to-end experience and identified collaboration opportunities.

IDEATION

Our First Idea - Flagging ( Did not ship )

As the only designer on the team, I began to ideate different solutions to this problem starting with low-fidelity and wire framing. The first first solution was Flagging.

SOLUTION HIGHTLIGHT

- Access to the platform

Once data scientists completed the analysis and review of the data, they made the data platform visible to partners. After access was granted, partners could navigate the dashboard and begin the flagging process.

- Streamlined Feedback Mechanism

The flagging process enabled partners to highlight data plots that needed revisions. They could use a free-text box to add context, including reasons for flagging the plots and instructions for the data scientists on how to address or modify the data.

- Real time and centralized communication

Partners and data scientists now communicate through a flagging modal, streamlining communication, improving alignment, and fostering better collaboration.

TESTING AND PROTOTYPING

Learnings and Feeback

The designs were presented to the partners to validate the proposed solution. These were they key areas that we uncivered during testing:

PRODUCT THINKING

A strategic reframe

Recognizing these challenges, I led the effort to reframe our strategy. Understanding that trust is earned through empowerment and transparency, I shifted our focus to empowering clients with control. Our new guiding question became:

“How can we empower clients through collaboration while building trust?”

IDEATION- FINAL SOLUTION

Edit Mode: Empowering Partners

After re-evaluating our approach to the problem , we embarked on a holistic redesign of the solution to empower customers with greater control over their data. This led to our shipped product; Edit Mode.

EARLY TOUCH POINTS FOR PARTNERS

CONTROLLED EDITING PERMISSIONS

SHARED ACCESS TO EDITING TOOLS

GENERATE RESULTS QUICKLY

FEEDBACK AND RESULTS

Internal Users

We launched the "Edit Mode" feature under controlled feature flags and rolled it out gradually to gather customer feedback and refine the user experience.
‍