
B2B SALES OPERATIONS
Moving a pump maker’s quotes out of Excel and into one tool that 100 people could share
ROLE
TIMELINE
TOOLS
FigmaTEAM
Anchal Nagdev (UX Designer), Atharva Ruikar (UX Designer), Rohit Mali (Product Manager), MID (development partner)
RESEARCH
Requirements mapping, then walkthroughs with Rohit and the sales team before each build. Nothing instrumented after launch.
Visuals on this page were recreated in Figma in 2026 from memory, since the original files weren’t retained.
THE CHALLENGE
A website couldn’t fix quotes that were built by hand in Excel
When I joined in January 2021 the ask was a website. Sales quoted out of spreadsheets, operations got orders with fields missing, and our map of that workflow moved the tool ahead of the site.
THE SOLUTION
One tool for the inquiry-to-handoff spine, a shared guide, and a website that feeds it
We scoped the tool to six stages: inquiry, requirements, product selection, quote, review, handoff. I designed inquiry capture and the quotation workspace, Atharva designed the information architecture and product selection, and one Figma guide gave both products the same forms and status labels.
~100
internal users
Sales, sales managers, and operations touched one record, so the handoff summary had to read the same to each.
6
quote states
Draft, needs information, under review, sent, accepted, closed. One label set for the tool, the guide, and the quote list.
5
requirement field states
Known, unknown, not applicable, needs follow-up, pending customer confirmation. A rep could start a quote without inventing a value.
3
product categories
Portable, vehicle-mounted, PTO. The website’s first entry path and the catalogue’s first layer.
THE PROBLEM
Reps quoted fast and operations paid for whatever was missing
A customer called. A rep typed the details into a sheet, found a pump in a PDF catalogue, and priced the quote by hand. The sheet couldn’t say which fields were still unknown, so reps typed a plausible value and moved on. The accepted quote reached operations as an emailed file, with phone calls to chase the gaps.
WHAT I OWNED
My half of a two-designer team
I owned the requirements mapping with Atharva, the inquiry capture screen, the quotation workspace, and the build reviews with MID. On the website I owned the entry paths, page structure, and responsive rules. In the guide I owned foundations, form components, and the content rules.
Atharva owned the tool’s information architecture, the catalogue, the selection and comparison flow, and the guide’s variants and library. He stayed 19 months after I left. Rohit Mali ran every walkthrough, and MID built from our Figma files and specs.
THE BRIEF I SET
How might we let a rep start a quote with what they know, and still hand operations an order with nothing missing?
Reps wanted the quote out the day the inquiry arrived. Operations needed every field filled before a pump was configured. My brief was one workflow that carried unknowns in the open, so both could be true.
THE CORE LOOP
Six stages, one record
A call, email, or web enquiry becomes one record. The technical fields can stay empty.
Each technical field takes one of 5 states, so nobody types a placeholder to pass a gate.
Atharva’s catalogue shows application fit first, then operating specs, then the full attribute set and documents.
Header, requirements, product, line items and terms, each block reading the one above it.
Bigger deals route through a manager, with notes and history under the pricing.
Operations gets an order with every unknown or pending field flagged.
Figure 1. The six stages as one record, growing a block at each stage, state chips colored the same way throughout.
DECISION 1
Let a field say “unknown” so nobody has to invent a value
Required fields looked like rigor and produced clean, wrong quotes. A rep who didn’t know the head or flow yet typed something plausible, and that number reached operations as a fact.
I gave every requirement field 5 states: known, unknown, not applicable, needs follow-up, pending customer confirmation. The inquiry screen asked for company, contact, site, and application first, and let the technical block sit in any mix of states. The same chips reappeared in the handoff summary, so a gap marked on day one stayed visible until the order left sales.

Figure 2. The inquiry capture frame, a state chip on each technical field, two of them set to “pending customer confirmation”.
DECISION 2
Lay the quote out in the order a rep works, lock the structure, leave the notes free
Excel had let every rep invent a layout, so a manager reviewed a different shape of quote from each of them.
I laid the workspace out top to bottom in the order a quote gets built. Header with customer, contact, owner, status, and next action. Then the requirement summary, the recommended product with its key specs, line items and terms, review notes, and last the handoff summary. Structure, product data, and status were locked. Commercial notes stayed editable, because negotiated terms didn’t fit a template and the sales team said so.
Six states carried a quote through the company: draft, needs information, under review, sent, accepted, closed. The status label, the next action, and the chip in the quote list came from one rule set in the guide.

Figure 3. The quotation workspace at desktop width, six blocks in work order, two open missing-information flags in the handoff summary.
DECISION 3
Build the tool before the website, and keep the first version to one spine
The business could’ve asked for a CRM or an ERP on top of the website brief. Our map put the failures in two places, information missing at capture and a handoff done by file and phone. A website couldn’t touch either, so I pushed the tool ahead of the site.
Version 1 stayed at the six stages. Inventory, reporting, a dealer portal, and service history waited. A smaller scope meant every part got a walkthrough with Rohit and the sales team before MID built it.

Figure 4. The requirements map, inquiry on the left and handoff on the right, the four deferred modules parked below the spine.
THE DESIGN GUIDE
One library, because two designers and an outside build team needed one set of rules
The tool and the website shared forms, tables, statuses, and product cards, so the guide came first, in six Figma layers: foundations, content rules, core components, product patterns, website modules, and documentation for MID. I owned foundations and form components and wrote the content rules. Atharva owned the variants and the library’s organization, and kept it current for 19 months after I left.

Figure 5. The six layers stacked, with the quote state chip pulled out to show one component serving both the tool and the site.
THE WEBSITE
Two ways in, one for buyers who know the model and one for buyers who only know the problem
The earlier site listed every product on one dense page. I rebuilt it around two entry paths. By category, portable, vehicle-mounted, or PTO, then family, then product detail. By application, for the buyer who knew the site conditions and nothing about model numbers.
Each product page layered application fit, key specs, technical documents, then the enquiry path. On a phone the spec table cut to the attributes that drive the decision, with the full table one tap away. The enquiry form reused the tool’s inquiry fields, so a web enquiry landed in sales with the same structure as a phone call.


Figure 6. A product detail page at desktop and phone width, the four layers labeled down the margin.
WHAT GOT BUILT
Eight modules on one guide, and no numbers to show for them
MID built the tool from our Figma files and specs and rolled it out to about 100 people in sales, sales management, and operations. The website went live after the tool. The current Firefly site is later work by other people, and nothing on this page describes it.
Nothing was instrumented, so there’s no inquiry-to-quote time, revision count, or adoption number here, and I won’t invent one. Validation was the walkthroughs with Rohit and the sales team before each build.
WHAT I’D MEASURE
What I’d measure, given a baseline
REFLECTION
What I’d redo
I’d set those measures in the first month, before any screen existed. We made every roadmap call on walkthrough feedback, and two years in we still couldn’t say how much faster a quote got. Even a weekly tally of call-backs in a spreadsheet would’ve given us a baseline.
I’d also put operations in the room from the first mapping session. The handoff summary was designed for them, and they saw it after sales did.
Project takeaways.
- Map the workflow before drawing a screen.Our first weeks of mapping turned a website brief into a tool brief and set the order of all four streams.Proficiency in requirements mapping
- Give the unknown a state of its own.Five field states let a rep save a half-known inquiry and let operations see what was still half-known.Proficiency in form and state design
- One guide for two products.The tool and the site shared forms, tables, and product cards, so one Figma library covered both.Proficiency in design systems
- Review the build like a user test.I checked each MID build against our frames before it reached sales. Where a build drifted, the fix went into the guide.Proficiency in developer handoff
