hero image: Bitsight VRM questionnaire interface, showing a question with answer options, score, and action rail
Case Study · Product Design

Redesigning the questionnaire that decides vendor risk

Product Design User Research Design Systems Accessibility
Timeline 2023-2024
My Role Senior Product Designer working across product strategy, research and design systems
Deliverables Questionnaire redesign, vertical action rail, scoring UI, PURE evaluation and usability test reports

After Bitsight acquired ThirdPartyTrust, the questionnaire feature needed more than a visual refresh. It needed a new foundation. I led the UI redesign and research plan for a complex, high-stakes workflow used by security teams every day.

1
1 reusable component from the Questionnaire, shipped to Bitsight DS and ready for reuse across all Bitsight products
5
5 independent question-level tools — review, bookmark, finding, messaging, internal notes — consolidated into one consistent action rail
100%
Task success rate in usability testing with existing users

A feature inherited from an acquisition

In Q3 2022, Bitsight acquired ThirdPartyTrust, a vendor risk management platform with its own established product and user base. As part of the integration, the Questionnaires feature needed to be rebuilt inside Bitsight's product ecosystem. That meant redesigning it from scratch using the Bitsight Design System, with better accessibility, clearer usability, and feature parity with the original.

Questionnaires are central to how security teams manage vendor risk. They come in three forms: custom questionnaires built by the team for specific assessments, internal questionnaires used for scoping during vendor intake and reassessment cycles, and industry-standard templates like CAIQ v4, ISO 27001:2022, and SIG Core.

This wasn't a greenfield project. It was a migration with real users watching. The design had to earn trust from people who already had a workflow, muscle memory, and strong opinions about how the tool should behave.

Feature parity isn't enough: it has to be better

The original platform had been built using its own framework, before any design system existed. Visual inconsistencies were everywhere, accessibility was not a priority, and some workflows required too many steps to accomplish basic tasks. At the same time, users had deep familiarity with the old interface. Every change needed a reason strong enough to justify the disruption.

The challenge had three parts. First, align the interface with Bitsight's design system without losing the features users depended on. Second, improve usability and accessibility across a complex, multi-part workflow. Third, do it iteratively: delivering features in planned cycles while the product was already in use.

Iterating toward the right answer

The design process started with a complete inventory of the existing platform, followed by the first designs and user flows built on the Bitsight DS. Each iteration was scoped to a set of features that could be shipped independently, which meant making early decisions about what to finalize and what to keep flexible as we learned more.

Feature inventory

Before starting to design, I ran a complete inventory of the questionnaire, every feature, interaction state, and edge case. Since there was no formal documentation, I did reverse engineering, mapping the whole system to understand what we needed to account for. This gave the team a shared map of what existed.

Feature inventory audit, old platform

The existing ThirdPartyTrust questionnaire interface, captured during the feature inventory audit before the redesign began.

Domain mapping

The feature inventory revealed something more than a list of things to build. It gave us a clear conceptual model from the reviewer's perspective, understanding which objects they were working with.

Domain mapping identified the core objects, their attributes, and the operations reviewers performed on them. It showed what existed, what was missing, and what relationships the new system had to support.

Domain model, deduced from the existing questionnaire

Domain-level object model, mapping the core entities, attributes, and relationships the new design had to account for.

The analysis revealed that Question is the unit of work, not the Questionnaire. Reviewers think and act at the question level. They flag, approve, annotate, and download one question at a time. The Questionnaire is a container; the Question is where decisions happen.

Questionnaire UI, light and dark mode side by side

Six core objects from the reviewer's perspective. The Questionnaire is a container; the Question is the unit of work.

Three distinct objects handle communication per question: Finding (visible to the vendor), Message (async communication with the vendor), and Internal Note (private to the review team, never shared). The domain model also shaped how the research tasks were written: tasks like "go to vendor X → open questionnaire Y → flag question N → create a finding" trace the exact path through the object hierarchy, which is part of why participants navigated the test intuitively.

The vertical rail

Each question in a questionnaire can have up to five tools associated with it: review (approve or flag), bookmark, finding creation, messaging, and internal notes. In the old platform, these were grouped in a confusing way and weren't identifiable at a glance.

I explored several layout approaches before landing on a solution. The tension was between discoverability (making tools visible enough that users knew they existed) and density (keeping the question interface readable when several questions are stacked in sequence).

Questionnaire UI, light and dark mode side by side

Three layout proposals for question-level tools. The vertical rail balanced visibility with density and became the foundation for the final design.

A vertical action rail anchored to each question kept tools consistently reachable without competing with question content. Actions are always in the same place, and users can scan a single vertical line to spot flagged questions or ones with findings.

Questionnaire UI, before and after

Before and after comparison of the questionnaire UI

Tool rail scrolling behavior

The scoring system

The scoring system calculates risk from three inputs: answer impact, question priority, and category weight, combined into an overall score. The result showed up as a small badge near the logo, but reviewers couldn't link it to what they were doing at the question level, or see how each action affected the score.

The solution was to show everything related to the score in one unified visual, which let reviewers intuitively understand what was driving the number.

The score uses a five-state system (Good, Fair, Warn, Bad, and N/A) adopted from the Bitsight DS, which had moved from a color-only scale to icon + color. This improved accessibility and kept the feature visually consistent with the rest of the product. At the category and questionnaire level the score becomes numerical, paired with the same icon, so reviewers can compare categories at a glance.

Questionnaire scoring system visualization

Score at question level (qualitative) vs. questionnaire level (numerical)

Two research studies

With the initial designs and user flows built on the Bitsight DS, I ran two research studies to validate the direction and surface friction before full delivery. The goal was to test with real users while also getting structured expert feedback early, catching issues at both ends of the process.

Study 1
5 internal participants
PURE Assessment Review

The ThirdPartyTrust workflows weren't familiar to Bitsight employees, so before testing with real users, we first ran a review with internal stakeholders. PURE (Practical Usability Rating by Experts) uses a defined set of criteria to surface usability issues before external testing, faster and cheaper than waiting for real users to find them.

Study 2
5 external, 1 internal participant
Moderated Usability Test

Sessions with existing users of the ThirdPartyTrust platform. Participants completed representative tasks using the new design and verbalized their reactions. Primary metrics: task success rate, time on task, error patterns, and qualitative feedback.

What we found

The usability test results were encouraging. Participants completed all tasks successfully and responded positively to the new interface. One of the most common reactions was around the question tools, which felt clearer and easier to reach than in the original product.

Top positives

  • All tasks completed successfully. Existing users navigated the redesigned interface without significant errors.
  • The UI felt more intuitive. The general reaction was that it felt "more clear and easy to use."
Questionnaire info is more clear now, I can see which questions need attention just by looking at the toolbar.
Usability test participant On the redesigned question layout

Top opportunities

  • Bulk document download. Users needed to download all documents from a questionnaire at once, but the system required them to do it one by one, per question.
  • Cross-category filtering. Categories were rendered client-side, which limited filtering to within each category. Users expected to filter across the full questionnaire, a structural constraint from engineering, not design.
  • User-built questionnaires. Users wanted to build questionnaires themselves. In the current state, questionnaires were created on demand by the Customer Success team, a bottleneck users found frustrating.
We download the files question by question, imagine doing that on a questionnaire with more than 100 questions
Usability test participant On document download
Filtering by categories is frustrating
Usability test participant On cross-category filtering

The last two pain points were highly requested, but low feasibility for engineering at that stage. They went into the backlog with clear rationale.

Prioritization matrix, impact vs. feasibility

Prioritization matrix showing impact versus feasibility

A complex feature, delivered in pieces that held together

The redesign shipped progressively across planned delivery cycles. Each iteration was tested or reviewed before the next began. Problems were caught early, not after everything was built.

Light and dark themes

Because this project ran in parallel with the design token adoption work, the questionnaire became one of the first features to be fully tokenized. Every component in the redesign referenced semantic tokens, which meant both light and dark mode were supported from the start, with no separate design files and no duplication.

Questionnaire UI, light and dark mode side by side

Light and dark mode, ready at launch. Both themes came with the token system decision.

Collaboration with the vendor-side team

The questionnaire feature has two sides: the reviewer experience (what security teams see) and the respondent experience (what vendors see). I was on the VRM (Vendor Risk Management) team, on the side of the companies doing the review, and collaborated closely with the TMH (Trust Management Hub) team, who owned the vendor-facing side. That collaboration worked because I'd worked on both products at ThirdPartyTrust, before the acquisition split the experiences into two. Keeping both sides consistent required constant alignment on shared components and tokens.

Questionnaire UI, categories panel for review and edit modes

Categories panel in review mode and edit mode.

Questionnaire UI, question cards for review and edit modes

Question cards in review mode and edit mode.

Action sheets and dialogs for question management

Action sheets and dialogs for question management.

Results

  • Two research studies, a complete feature redesign, and a set of reusable components were delivered.
  • The usability test validated the core direction: all tasks were completed successfully, and participants responded positively to the new tool layout.
  • The pain points surfaced in research were documented, triaged, and communicated to the product team with clear feasibility context.

The questionnaire redesign in action

Customer pilot program

After launching to production, the product team ran a pilot program with both new and existing customers during the platform's unification phase. Legacy customers accessed VRM Beta for a preview and feedback gathering. New customers enrolled directly in the VRM Beta Pilot Program.

  • Confirm Bitsight VRM's value proposition by asking users whether the product was delivering according to their expectations
  • Test and learn for continuous improvement: gathering feedback on usability, functionality, and overall experience
  • Continue promoting Bitsight VRM to new customers while existing customers remained on the legacy platform
We need more flexibility on who we can assign a questionnaire or a question to
VRM Beta Pilot participant On recipient flexibility
You can use SAP Ariba or other TPRM products to send questionnaires easily but not necessarily create this kind of live experience… in real time you can start conversations, send messages to the supplier and cooperate having everything in one place
VRM Beta Pilot participant On real-time collaboration
What's amazing is this ability to easily define requirements and create these conversations to follow up about the questions
VRM Beta Pilot participant On question-level messaging

I didn't run the pilot program, but having that kind of feedback from a live platform, from real customers using the feature in production, was a meaningful signal that the design direction held up beyond the controlled conditions of usability testing.

What this project taught me

I learned how expensive it is to work without documentation. Building the feature inventory meant reverse engineering the whole product from scratch, and that took a real chunk of research time. Now I try to document as I go, even when nobody asks for it.

Running the PURE evaluation first caught issues before any user ever saw the design. It was the first time I ran this kind of evaluation, and I was surprised by how useful, fast, and cheap it turned out to be.

The biggest improvements users asked for were tied to custom questionnaires and questionnaire-level filtering. Both got scoped out of this phase, but the conversation about what to prioritize kept coming back throughout the project, and it taught me how to balance user experience against engineering feasibility.