Felipe Cruz

B2B Quotation Platform

I joined after the first releases, working alongside the Design Lead as the product evolved through an 18-month roadmap. Priorities were shaped by customer needs, issue severity and engineering effort. The quotation workflow was reduced to three steps, cutting average proposal time from 60 minutes to 15 and new-agent training from 12 hours to 4 hours.

Outcomes
60 min → 15 min average proposal time 12 hours → 4 hours new-agent training
Role
Research · Interaction Design
Area
Telecom · B2B Sales
Scale
500 sales agents
Timeline
18 months
Sales workspace with a product catalog beside an open quote and an Add action ready

Where I came in

Sales agents in a multinational telecommunications operator’s Brazil operation built one proposal across seven tools. They quoted small, mid-size and large companies across the country. Each service category had its own quotation system. Contracts, approvals, and orders lived somewhere else. Agents handled 20 to 30 customer interactions a day, and a quote could take an hour. I joined after the first releases, beside the Design Lead, for the 18 months that followed. My job was to decide which gaps were severe enough to ship, and to shape that roadmap with research and design.

The decision

Agents built one quote across seven tools. There was so much flexibility that a new agent could not tell where to start. Approval lived in side conversations, outside the quote. On a deal with several sites, the same fields were typed again for every site.

A quote took an hour because of that. We scored what we found by how often it happened, how much it hurt, and whether it stuck. Severe problems that were cheap to fix shipped first, and the long list of other opportunities waited. When the backlog asked for another screen, the Design Lead and I kept the product from growing a new feature. We were aiming to simplify things most of all.

Usability findings scored by frequency, impact and persistence, with a severity decision tree

We reorganized the quote into three stages: Find Product, Configure Solution, Generate Contract. Approval status and progress stayed visible inside that path, so agents stopped leaving the workspace to ask where a deal stood.

Two ways of selling kept showing up. Express quotes, done while the customer waited, with no customization. Custom deals, which could sit for approval for more specific scenarios. The three stages had to carry both. Averaging them into one happy path would have broken the slower deal.

Two ordering personas: simple express quoting versus complex customized deals that wait for approval
Before and after diagram of the sales workflow: seven fragmented tools condensed into four calm steps, cutting quote time from sixty to fifteen minutes and onboarding from twelve hours to four hours.

The quote

Find Product is a catalog beside the open quote. Search, filter, add, without leaving the deal.

Catalog panel beside an open quote, with a product selected and an Add button ready to place the offer into the cart

Configure Solution covers multi-site deals. Several sites, one side panel, shared term, technology, and billing rules. The agent stops retyping the same fields per site.

Quote workspace with a configuration panel open: term, technology, channels, national tariffs and SLOs, then Calculate Price

Renewals sat in the same place. Search the install base, select what changes, decide how a bundle is renegotiated, then continue the quote.

Install Base panel listing active customer services with multiple items selected for modification
Combo Decision step where the agent chooses how to renegotiate selected services in a bundled offering

Results

60 min → 15 min

A quote fell from about an hour to about fifteen minutes on average, customized quotes included. Fifteen minutes was the average, not a time every quote had to hit. The agent stayed in one workspace instead of moving across seven tools.

12 hours → 4 hours

New agents had trained across three mornings, four hours each. That became one morning. Twelve hours became four, and the path was easier to learn.

Experienced agents used the platform for three months before it spread across the organization. Those three months were there to see whether it held in every scenario, to find bugs, and to find improvements that could enter the backlog.