Home

Redesigning the sourcing-to-quoting pipeline to drive product adoption

AT A GLANCE

Two design bets closed the sourcing-to-quoting gap, boosting conversion by 50% across 50+ clients

Luminovo builds software that helps electronics manufacturers (OEMs) find suppliers, compare offers, and get quotes for their components. When I joined, the platform was losing users at two points: sourcing was slow because key information was buried in dense screens, and quoting still happened outside the platform, through email and spreadsheets, which meant platform data was always out of date.

My team and I owned the end-to-end redesign of the sourcing-to-quoting flow and made two bets:

  • Redesigned the sourcing experience from the component overview to the component details screen where users compare suppliers for each one.

  • Built quoting directly into the platform, replacing a slow and manual process over email and phone. This meant users no longer had to leave the platform to get supplier quotes, everything now lived in one place.

Role

Product Designer

Company

Luminovo

Timeline

8 months

Team

1 Product Manager, 2 Product Engineers, and 1 Founder

Tools

Figma, FigJam, Token Studio, Amplitude, LogRocket, Notion

50% increase

sourcing completion rate

~60% faster

quotation cycle

50+ clients

benefited from redesign

before
after
before
after

before/after comparison of the solution manager

CHALLENGE

Sourcing was too slow, and quoting happened entirely off-platform — creating a hard ceiling for product adoption

OEMs using LumiQuote needed to find the right electronics manufacturing suppliers (EMS) for their bill of materials, compare offers across vendors, and get quotes. In practice, both halves of this process were broken.


  • Sourcing was slow and frustrating: The interface buried the information OEMs needed under dense, confusing layouts:

  • Overwhelming information density: The aggregated BOM overview and per-component solution manager stacked scenario overviews and offer descriptions together. When a part had 10-20+ offers, there was no way to filter or rank them.

  • Manual, inefficient workflows: No filters existed. Users had to scroll through every offer manually to find the best match.

  • Disorienting navigation: Without a global navigation bar or clear column labels, users struggled to track their context or understand data points.

  • Unpolished interface: Irregular spacing and inconsistent layout scales made the dense screens even harder to scan.

These issues meant OEMs spent far more time than necessary finding and evaluating the right suppliers, which directly hurt the platform's conversion rate.


Quoting left the platform entirely

The workflow ran in three phases: finalize solutions and select parts inside LumiQuote, then email the supplier with a quote template and wait for a reply (often in a different format), then manually add the quote offers back into LumiQuote and fill missing fields. The quote template was complicated for suppliers. The import couldn't handle different Excel versions. The export was manual and couldn't flag unquoted parts. This loop kept platform data perpetually stale, turnaround times unpredictable, and blocked meaningful product adoption.

Neither problem surfaced clearly in customer feedback. Users couldn't articulate the root cause. It took a combination of analytics and user interviews to make the structural break legible.


CHALLENGE

Sourcing was too slow, and quoting happened entirely off-platform — creating a hard ceiling for product adoption

OEMs using LumiQuote needed to find the right electronics manufacturing suppliers (EMS) for their bill of materials, compare offers across vendors, and get quotes. In practice, both halves of this process were broken.


  • Sourcing was slow and frustrating: The interface buried the information OEMs needed under dense, confusing layouts:

  • Overwhelming information density: The aggregated BOM overview and per-component solution manager stacked scenario overviews and offer descriptions together. When a part had 10-20+ offers, there was no way to filter or rank them.

  • Manual, inefficient workflows: No filters existed. Users had to scroll through every offer manually to find the best match.

  • Disorienting navigation: Without a global navigation bar or clear column labels, users struggled to track their context or understand data points.

  • Unpolished interface: Irregular spacing and inconsistent layout scales made the dense screens even harder to scan.

These issues meant OEMs spent far more time than necessary finding and evaluating the right suppliers, which directly hurt the platform's conversion rate.


Quoting left the platform entirely

The workflow ran in three phases: finalize solutions and select parts inside LumiQuote, then email the supplier with a quote template and wait for a reply (often in a different format), then manually add the quote offers back into LumiQuote and fill missing fields. The quote template was complicated for suppliers. The import couldn't handle different Excel versions. The export was manual and couldn't flag unquoted parts. This loop kept platform data perpetually stale, turnaround times unpredictable, and blocked meaningful product adoption.

Neither problem surfaced clearly in customer feedback. Users couldn't articulate the root cause. It took a combination of analytics and user interviews to make the structural break legible.


Analytics and user interviews revealed two hidden roadblocks: an overloaded sourcing interface that slowed decision-making, and an off-platform quoting loop that kept the system incomplete.

SOLUTION

Redesigning the full sourcing-to-quoting flow, backed by a scalable design library

The work addressed both halves of the broken pipeline. I redesigned the sourcing experience end-to-end: a new aggregated BOM overview for cross-component status, and a rebuilt solution manager for per-component supplier decisions. In parallel, I designed an automated quoting module that replaced the email loop and made LumiQuote the system of record. A shared design library ran alongside the product work, keeping delivery fast and consistent.

SOLUTION

Redesigning the full sourcing-to-quoting flow, backed by a scalable design library

The work addressed both halves of the broken pipeline. I redesigned the sourcing experience end-to-end: a new aggregated BOM overview for cross-component status, and a rebuilt solution manager for per-component supplier decisions. In parallel, I designed an automated quoting module that replaced the email loop and made LumiQuote the system of record. A shared design library ran alongside the product work, keeping delivery fast and consistent.

Aggregated BOM overview

Redesigned the top-level bill of materials view to give OEMs a clear, scannable overview of sourcing status, prices, and selected suppliers across all components. Added sorting and filtering so users could quickly identify which parts still needed attention, replacing a dense screen that offered no way to prioritize.

Automated in-platform quoting

Designed Quote Price 2.0 to replace the email-and-spreadsheet quoting loop. OEMs could configure scenarios, send bulk RFQs, and import quotes in any Excel or PDF format. Suppliers uploaded responses directly. Six configurable views and a smart line card with automatic supplier suggestions made the process self-contained.

Rebuilt solution manager

Redesigned the per-component screen where OEMs compare supplier offers and make sourcing decisions. Added a global navigation bar, clear breadcrumbs, dedicated filters, and a sortable offer table surfacing the top three options (best price, cheapest, fastest). Hidden the scenario overview by default to reduce cognitive load.

Design library foundation

Led the structural optimization of Luminovo's shared Figma design library, addressing architectural inconsistencies through audits and standardized usage guidelines. Managed the full component lifecycle from backlog definition and competitive research to stakeholder reviews and technical validation. Deployed four core redesigned components to production, including a complex Code Editor.

Aggregated BOM overview

Redesigned the top-level bill of materials view to give OEMs a clear, scannable overview of sourcing status, prices, and selected suppliers across all components. Added sorting and filtering so users could quickly identify which parts still needed attention, replacing a dense screen that offered no way to prioritize.

Rebuilt solution manager

Redesigned the per-component screen where OEMs compare supplier offers and make sourcing decisions. Added a global navigation bar, clear breadcrumbs, dedicated filters, and a sortable offer table surfacing the top three options (best price, cheapest, fastest). Hidden the scenario overview by default to reduce cognitive load.

Automated in-platform quoting

Designed Quote Price 2.0 to replace the email-and-spreadsheet quoting loop. OEMs could configure scenarios, send bulk RFQs, and import quotes in any Excel or PDF format. Suppliers uploaded responses directly. Six configurable views and a smart line card with automatic supplier suggestions made the process self-contained.

Design library foundation

Led the structural optimization of Luminovo's shared Figma design library, addressing architectural inconsistencies through audits and standardized usage guidelines. Managed the full component lifecycle from backlog definition and competitive research to stakeholder reviews and technical validation. Deployed four core redesigned components to production, including a complex Code Editor.

Solution deep dives

Redesigning the sourcing experience from BOM overview to solution manager

The sourcing flow had two layers, and both were failing. The aggregated BOM overview gave no clear picture of sourcing status across components. The per-component solution manager buried supplier offers under dense, unfiltered screens. A heuristic audit had named the violations: cognitive load, absent filters, recall over recognition, and inconsistent layout standards. The harder question was sequencing: what to fix first, and why.

I triangulated before touching the design. Amplitude funnel data showed high drop-off rates at the solution manager entry point. LogRocket session recordings made the failure concrete: users paused, clicked randomly, and went inactive before abandoning. User interviews confirmed the audit hypothesis and surfaced a structural need: OEMs needed to operate at two levels simultaneously, a BOM-level view to track status across all components and a per-component view to fine-tune supplier selections. Serving both in one undifferentiated screen was the root of the cognitive load problem.

The redesign split the experience into two connected levels. The aggregated BOM overview was rebuilt to show sourcing status, prices, and selected suppliers at a glance, with sorting and filtering to help users prioritize which parts still needed attention. The solution manager was rebuilt around the per-component decision: a global navigation bar replaced URL-based navigation, breadcrumbs and a heading with total offer count replaced the opaque "Part options" label, the scenario overview was hidden by default to surface only the best solution, and a sortable table with filters replaced manual scrolling through 10-20+ offers. Three iteration rounds moved from internal testing during live customer calls to formal sessions with the OEM users who had flagged the original problems. After production rollout, Amplitude data showed the sourcing completion rate had doubled.

The redesigned sourcing flow: aggregated BOM overview for cross-component status, and a rebuilt solution manager with navigation, filters, and sortable offer table.

Redesigning the sourcing experience from BOM overview to solution manager

The sourcing flow had two layers, and both were failing. The aggregated BOM overview gave no clear picture of sourcing status across components. The per-component solution manager buried supplier offers under dense, unfiltered screens. A heuristic audit had named the violations: cognitive load, absent filters, recall over recognition, and inconsistent layout standards. The harder question was sequencing: what to fix first, and why.

I triangulated before touching the design. Amplitude funnel data showed high drop-off rates at the solution manager entry point. LogRocket session recordings made the failure concrete: users paused, clicked randomly, and went inactive before abandoning. User interviews confirmed the audit hypothesis and surfaced a structural need: OEMs needed to operate at two levels simultaneously, a BOM-level view to track status across all components and a per-component view to fine-tune supplier selections. Serving both in one undifferentiated screen was the root of the cognitive load problem.

The redesign split the experience into two connected levels. The aggregated BOM overview was rebuilt to show sourcing status, prices, and selected suppliers at a glance, with sorting and filtering to help users prioritize which parts still needed attention. The solution manager was rebuilt around the per-component decision: a global navigation bar replaced URL-based navigation, breadcrumbs and a heading with total offer count replaced the opaque "Part options" label, the scenario overview was hidden by default to surface only the best solution, and a sortable table with filters replaced manual scrolling through 10-20+ offers. Three iteration rounds moved from internal testing during live customer calls to formal sessions with the OEM users who had flagged the original problems. After production rollout, Amplitude data showed the sourcing completion rate had doubled.

The redesigned sourcing flow: aggregated BOM overview for cross-component status, and a rebuilt solution manager with navigation, filters, and sortable offer table.

Bringing quoting in-platform to close the sourcing loop


The biggest drop-off in the LumiQuote funnel wasn't in the UI — it was after it. OEMs would complete the sourcing phase, then leave the platform to manage RFQs through email, spreadsheets, and phone calls. Turnaround times were long and variable. Platform data was always behind. The sourcing flow was technically complete but the workflow was broken.

User interviews surfaced this explicitly: customers couldn't articulate the quoting process as a platform failure because they had normalized the off-platform workaround. But when asked about their biggest time costs, the quote cycle came up immediately. Amplitude data showed a corresponding pattern: sourcing completions that never converted to downstream platform activity, indicating the workflow was ending at the point of quoting.

The decision was to build Quote Price 2.0 module — not a minor improvement to the existing flow, but a new capability. The module let OEMs configure quoting scenarios across multiple suppliers, adjust variables, and send RFQs directly from the platform. Supplier responses were tracked in-platform, keeping data current and replacing the email thread as the source of truth. The design went through multiple prototype rounds, tested first with internal colleagues during customer calls and then with the OEM users who had described the off-platform problem most clearly. The phased rollout — select customers first, broader release later — let us validate before full production. The estimated turnaround time reduction was approximately 60%, based on team observation of before-and-after quoting cycles.

Quote Price 2.0: redesigned tables, quick selection, smart line card, robust import, bulk email, and six scenario views replaced the three-phase email-and-spreadsheet quoting loop.

Bringing quoting in-platform to close the sourcing loop


The biggest drop-off in the LumiQuote funnel wasn't in the UI — it was after it. OEMs would complete the sourcing phase, then leave the platform to manage RFQs through email, spreadsheets, and phone calls. Turnaround times were long and variable. Platform data was always behind. The sourcing flow was technically complete but the workflow was broken.

User interviews surfaced this explicitly: customers couldn't articulate the quoting process as a platform failure because they had normalized the off-platform workaround. But when asked about their biggest time costs, the quote cycle came up immediately. Amplitude data showed a corresponding pattern: sourcing completions that never converted to downstream platform activity, indicating the workflow was ending at the point of quoting.

The decision was to build Quote Price 2.0 module — not a minor improvement to the existing flow, but a new capability. The module let OEMs configure quoting scenarios across multiple suppliers, adjust variables, and send RFQs directly from the platform. Supplier responses were tracked in-platform, keeping data current and replacing the email thread as the source of truth. The design went through multiple prototype rounds, tested first with internal colleagues during customer calls and then with the OEM users who had described the off-platform problem most clearly. The phased rollout — select customers first, broader release later — let us validate before full production. The estimated turnaround time reduction was approximately 60%, based on team observation of before-and-after quoting cycles.

Quote Price 2.0: redesigned tables, quick selection, smart line card, robust import, bulk email, and six scenario views replaced the three-phase email-and-spreadsheet quoting loop.

Design system as delivery infrastructure

The product work moved fast — four modules redesigned across seven months, with new components entering production on a rolling basis. Without a shared component foundation, every design-to-engineering handoff risked inconsistency and rework.

I joined Luminovo's shared Figma design library as a contributor alongside engineers and designers, using Token Studio to establish a structured token system that could be maintained as the product evolved. This wasn't a separate initiative — it ran in parallel with the LumiQuote work and fed directly into it. Every new component designed for the solution manager or quoting module was built to library standards and documented for developer handoff.

The most complex output was a redesigned Code Editor component, one of four core components deployed to production. It required close collaboration with engineers to resolve implementation constraints while preserving the interaction design intent. The Notion-based backlog tracked component lifecycle from audit through production validation. The result was a design-development pipeline that could handle rapid iteration without accumulating inconsistency — a foundation that set the baseline for Luminovo's future UI scalability.

Four redesigned components deployed to production, including the Code Editor — showing the design-to-code pipeline that enabled fast, consistent delivery.

Design system as delivery infrastructure

The product work moved fast — four modules redesigned across seven months, with new components entering production on a rolling basis. Without a shared component foundation, every design-to-engineering handoff risked inconsistency and rework.

I joined Luminovo's shared Figma design library as a contributor alongside engineers and designers, using Token Studio to establish a structured token system that could be maintained as the product evolved. This wasn't a separate initiative — it ran in parallel with the LumiQuote work and fed directly into it. Every new component designed for the solution manager or quoting module was built to library standards and documented for developer handoff.

The most complex output was a redesigned Code Editor component, one of four core components deployed to production. It required close collaboration with engineers to resolve implementation constraints while preserving the interaction design intent. The Notion-based backlog tracked component lifecycle from audit through production validation. The result was a design-development pipeline that could handle rapid iteration without accumulating inconsistency — a foundation that set the baseline for Luminovo's future UI scalability.

Four redesigned components deployed to production, including the Code Editor — showing the design-to-code pipeline that enabled fast, consistent delivery.

The Process

How the work unfolded

Step 1: UX audit and domain immersion

I began with a heuristic evaluation, triggered by a platform resolution update that exposed latent layout issues. The audit identified high cognitive load, absent filters, and navigation flaws. Simultaneously, I immersed myself in the supply-chain process. This domain knowledge turned a list of UI problems into a structural diagnosis of how OEMs evaluate BOMs.


Step 2: Triangulating evidence

Audit findings alone couldn't justify a major redesign. I cross-referenced them with Amplitude funnel data (showing drop-offs) and LogRocket replays (showing inactivity and abrupt abandonment). After finding the same failures across three sources, we interviewed struggling OEMs. This confirmed our hypotheses and exposed the off-platform quoting loop, cementing the project scope.


Step 3: Design, prototype, and internal validation

I designed multiple iterations of the end-to-end flow. The BOM overview was rebuilt for cross-component status, the solution manager gained dedicated filters, and Quote Price 2.0 was designed to bring workflows in-platform. We tested prototypes internally—having colleagues use them on live customer calls—which drove two major iterations before formal testing.


Step 4: User testing and phased rollout

We ran testing sessions with the OEMs who originally flagged the problems, leading to targeted pre-production refinements. We used a phased rollout, releasing to select customers first to safely monitor Amplitude data. Full release followed only after metrics proved success: sourcing flow completions had doubled.


Step 5: Design library in parallel

Throughout the project, I contributed to Luminovo's shared Figma library. I conducted audits, defined usage guidelines, and collaborated with engineers to deploy four core components, including a complex Code Editor. This parallel system work ensured our rapid iteration cycles remained consistent and sustainable.


Note: Due to NDA obligations, I cannot share all research details or behind-the-scenes screenshots.

Gallery

Reflections

01

Clearly articulating a problem is hard

Most LumiQuote users didn't complain about the quoting loop, but requested peripheral features instead. I learned to treat customer complaints as weak signals and use analytics and interviews to surface what users can't articulate.

03

Shared purpose beats time zones

Despite never being in the same room, the fully remote and multicultural team trusted each other's expertise and moved quickly. That trust made it easier to advocate for disruptive scope changes because we were oriented around the same problem, not the same geography.

02

No single data point tells the full story

Amplitude showed where users dropped off, LogRocket showed what they did, and interviews explained why. This triangulated process became my default approach: start with funnel data, inspect session replays, and then talk to users at the point of failure.

04

Domain immersion rewards well

Once I understood how OEMs think about BOMs and RFQ cycles, every design decision became sharper and easier to defend. I now treat domain immersion as an ongoing design deliverable that keeps improving the work, rather than just a prerequisite.

01

Clearly articulating a problem is hard

Most LumiQuote users didn't complain about the quoting loop, but requested peripheral features instead. I learned to treat customer complaints as weak signals and use analytics and interviews to surface what users can't articulate.

02

No single data point tells the full story

Amplitude showed where users dropped off, LogRocket showed what they did, and interviews explained why. This triangulated process became my default approach: start with funnel data, inspect session replays, and then talk to users at the point of failure.

03

Shared purpose beats time zones

Despite never being in the same room, the fully remote and multicultural team trusted each other's expertise and moved quickly. That trust made it easier to advocate for disruptive scope changes because we were oriented around the same problem, not the same geography.

04

Domain immersion rewards well

Once I understood how OEMs think about BOMs and RFQ cycles, every design decision became sharper and easier to defend. I now treat domain immersion as an ongoing design deliverable that keeps improving the work, rather than just a prerequisite.

Luminovo wasn't just remote-first, it was borderless in spirit. Working alongside a multicultural team with English as our common thread, I learned that great ideas don't need a shared hometown to take shape. Once a luminerd, always a luminerd.

‹ Designing white-label infrastructure for private markets