XBase
If we met in an elevator, this is what I’d tell you.
As Flexe’s head of design, with no PM, I led the vision, design, and rollout of XBase: one platform that took every business process out of people’s heads and handoff meetings and into software. It started as a configuration tool and became the front end for the whole company. Errors fell 99%, net margin went from -2.5% to +12%, and engineers got 1,200 hours a month back.
A guided, TurboTax-style flow let operations teams answer plain questions while XBase built the configuration file.
Templates for every program and every enterprise customer turned each new site into a few clicks.
Account managers kicked off a launch mid-call and assigned each module to the right person across the company.
Account managers got one dense page so calls felt like conversations, and every other team got short, focused steps so nothing was missed.
Dependent settings appear on their own, formatting is handled for you, and every configuration runs on mock data before go-live.
Executives saw the whole pipeline and every account’s numbers, with big swings flagged automatically.
Every internal tool behind one login, with each person seeing only their customers and their lane.
Growing pains
A brand like Casper sells two mattresses in three sizes. Enterprise retailers like Walmart, Nike, and Home Depot ship hundreds of thousands of products, thousands of units a day, out of tight spaces. Every new deal brought more steps, more requirements, and more people to keep track of it all. In about 18 months, we went from a 15% margin to losing money on every new account, leaving Flexe at a -2.5% net margin.
3–5accounts per account manager
15%+billing errors
1,200engineering hours a month on configuration
15% →−2.5%net margin
The messy middle
To cut costs, our Director of Technical Solutions wanted every customer to follow the same path, passed from team to team, with requirements collected in one spreadsheet. She asked me to map our intake-to-launch process. The map showed how messy it really was, and that passing work team to team would be far too slow. The spreadsheet showed who owned each step, but it couldn’t connect the dependencies between teams, and that’s where the errors occurred.
The first pitch
Even with the spreadsheet, engineers still had to hand-code every configuration. So I asked: why have engineers doing this at all? Why not empower our teams to do the configurations themselves? With the right tool, we could also track the dependencies automatically, and those were causing our errors. I ran through a sample demo of facility requirements. Our CEO’s reaction: in our wildest dreams, but far too complex and costly to build, with a roadmap already full.
Lesson learned · Lead with the business problem, not the tool.
Building a coalition
After my first pitch, I worked with our VP of Operations to form a coalition of leaders across business and operations, and kept building a more functional version of the prototype.
Round two
This time, I didn’t open with the prototype. I opened with what success would look like for every team.
⚑ Not written yet: what success looks like for each team
Team name
Success line
Team name
Success line
Team name
Success line
Team name
Success line
Team name
Success line
Team name
Success line
Proof of concept
The first official project: a simple Shopify customer at an existing site. Built from scratch in XBase, it was as fast as an engineer copying a matching customer’s files, with no engineer involved.
Under one roof
Configuration, reservations, accounts, billing, Marketplace, and more, all in one place.
Role-based access meant each person saw only their own customers and lane, which got us SOC 2 compliant for enterprise customers.
No loose ends
The outcome
3–5 →10–15Accounts per account manager
99%fewer billing and configuration errors
1,200engineering hours a month, back to building product
−2.5% →+12%Net margin
