B2B Client Dashboard @Brarista
Role
Lead Product Designer
Responsibilities
UX Research
UX/UI Design
Usability Testing
Dev Handoff
Team
Tools
Figma
Mongo DB
Notion
Clickup
BACKGROUND
Too Many Tabs! 😵💫
Brarista is the world's first conversational AI platform built specifically for lingerie retail. By combining intelligent fit technology with professional fitting expertise, it tackles one of the industry’s most persistent and costly problems: 80% of women wear an ill-fitting bra. As a B2B2C white-label solution, Brarista helps lingerie retailers reduce returns, increase shopper confidence, and improve overall profitability.
As Brarista gained traction, managing B2B client conversations and requests began to scatter across WhatsApp threads, Trello, and ClickUp. Wearing multiple hats in a fast-growing start-up, I began to notice the inefficiency of fragmentation and the quiet erosion of client trust that came with it.
This case study walks through how I led the design of a centralised B2B dashboard that stemmed from this internal pain point. I drove the design direction through discovery to MVP (and now into its next iteration), and coordinated cross-functional delivery between design and engineering.
This case study focuses on the design process and learnings; specific UI details remain under NDA.
Creativity is messy. An adapted Double Diamond process validated ideas early and reduced wasted development time.
PROBLEM
Dissecting and Comprehending the Problem 🤓
We initially assumed the key challenge was streamlining client requests. It sounded reasonable, yet it remained an inside-out assumption, shaped more by our internal pain than by the clients’ lived experience.
Understanding the problem is the foundation of any meaningful solution. To ground it in reality, we used a mixed-method approach:
Client Interviews
Two clients, one long-term partner familiar with our system and one new client still navigating onboarding. This contrast helped us understand both ends of the adoption curve.
Internal Brainstorming
Collaborative sessions with the founder and engineers to surface operational bottlenecks.
Data Analytics
A review of chatbot logs and performance metrics to assess what patterns were visible and what was technically attainable.
Each interview followed principles from The Mom Test, asking specific, past-oriented questions instead of hypothetical ones. I listened more than I spoke, looking for behavioural truth rather than surface-level preference.
Sneak peek into interview guide in Notion. Confidential information is blurred.
Research revealed a hidden pain point, and reframed the problem entirely.
While streamlining requests mattered, clients’ deeper frustration was the lack of visibility. There was a fog around how the chatbot actually performed and what stories lived within the data. Even with regular monthly catch-ups, clients found it difficult to see beneath the surface of their own AI assistant.
As we analysed our chatbot data more closely, we realised that hidden within the chatbot logs were numeric traces of consumer behaviour that revealed the latent potential of a B2B2C feedback loop. Our B2B clients could, in fact, benefit directly from these downstream insights about their end consumers.
In the lingerie industry, an ecosystem defined by intimacy and emotional resonance, these numbers told human stories. Conversations with end consumers carried cues about fit issues, size distributions, and emerging shopper archetypes that could guide everything from marketing strategy to product development for our clients.
Ultimately, we understood that our clients weren't asking us to erase human connection with automation. They wanted to hear the conversations through data and understand their consumers at scale.
This reframing changed how we understood success. It was no longer just about smoother internal operations; it became more about restoring confidence and decision-making power to our clients. In essence, the problem evolved from a technical limitation to a question of trust and transparency.
From these insights, we defined three guiding hypotheses to steer the product vision:
Visibility & Trust
Providing clients with real-time, intuitive access to chatbot performance metrics would strengthen transparency and reinforce confidence in our service.
Actionable Insight
Turning conversational data into lingerie-specific behavioural and demographic insights would help clients make smarter business decisions and position Brarista as a strategic partner rather than a vendor.
Operational Cohesion
Centralising client communication and analytics would reduce friction, accelerate response times, and create a more cohesive client experience.
These hypotheses became the foundation of the dashboard’s information architecture, placing metrics first and requests second.
DESIGN
Agile, Iterative Design 🧐
The design process didn't follow a rigid, linear path. Instead, it unfolded through agile cycles of iteration, prototyping, testing, and refinement. Concepts were drafted and prototyped in Figma, tested internally and with select clients, then reshaped through rapid feedback loops. This iterative approach allowed ideas to mature organically while staying grounded in real user and business needs.
To balance agility with structure, I implemented an agile user story framework supported by the MoSCoW prioritisation method. This clarified what needed to be built first and what could evolve later, aligning design scope with engineering capacity.
Sneak peek into user story documentation on ClickUp
The handoff was not a single moment of transfer, but a phase of co-creation between design and engineering.
Before development began, user stories were finalised to articulate each feature’s purpose, functionality, and measurable outcome. These were linked directly to an interactive Figma prototype, creating a shared source of truth for the team.
With engineers working remotely (and a 6 hour time difference), I prioritised clarity over formality. Depending on context, I used annotated Figma files, short video walkthroughs, or live co-working sessions to clarify interactions in real time. Quick calls often proved more effective than tickets for resolving nuanced design logic or fine-tuning micro-interactions.
QA followed the same iterative cadence. Each dashboard build was reviewed against predefined acceptance criteria and approved once both UX and technical performance met expectations.
The result was an MVP that aligned client expectations with Brarista’s strategic objectives. Although specific UI details remain confidential, the dashboard’s first release included the following key capabilities:
Real-Time Visibility into Performance
Engagement summaries and conversion analytics
ROI and attributed order tracking
Transparent access to full conversation logs
Exposure to Behavioural Insights
Thematic conversation categorisation
Detailed size distribution and fit-related analytics
Insights into bra-wearing habits
Streamlined Operational Communication
Centralised dashboard integrating requests, data, and updates in one coherent space
LEARNINGS
Midway Takeaways 🤠
This project unfolded as both a design challenge and a systems lesson. Building Brarista’s B2B dashboard meant navigating not just user experience, but collaboration that make design operationally real.
Never Assume
What we think clients want and what they actually need are rarely identical. Early assumptions around efficiency evolved into insights about visibility and trust. This required continuous alignment across all stakeholders (i.e. clients, founder, and engineers) to ensure priorities reflected both user value and business reality.
Good Design Requires Good Rapport
No amount of documentation replaces human connection. The closer I worked with developers, the smoother and more meaningful the handoff became. Live QA reviews built shared understanding far faster than long written specs.
(And yes — becoming friends with your engineers genuinely helps.)
Systems Thinking Strengthens Design Outcomes
Having foundational knowledge in software engineering shaped how I structured user stories, prioritised development scope, and defined acceptance criteria. This hybrid lens between design and systems thinking allowed for a product that was not just desirable, but buildable.
