Case study · Cleaning · commercial + residential
Shine and Sparkle
Three AI workflows freed 40 hours a month for a cleaning business — a different industry, the same playbook.
Case study · Unified Fleet Services
From a full-day quote to minutes.
John Scott is CEO of Unified Fleet Services, a fleet equipment leasing company serving government and non-profit clients. The company was capping its growth on a manual, spreadsheet-driven quoting process, and two data analysts spent their days wrangling templates instead of analyzing anything. Here's how we replaced it — and what changed when the bottleneck went away.
- Client
- Unified Fleet Services · John Scott, CEO
- Industry
- Equipment leasing · government + non-profit
1 day → minutes
Quote turnaround time
277
Quotes generated in the first month
2 FTEs
Data analysts freed from spreadsheet wrangling
The challenge
Spreadsheet-driven quoting was capping the entire business.
Quoting in fleet leasing is unusually fluid: interest rates change, vehicle depreciation values shift, every state has its own tax treatment, and every customer wants to see multiple term and mileage scenarios before committing. Unified Fleet had built that complexity into a sprawling library of spreadsheets, with two full-time data analysts holding the whole thing together.
A spreadsheet template per scenario
Every mileage option had its own spreadsheet template. Every vehicle depreciation lived in a separate sheet. Every interest rate scenario was its own tab. Two data analysts were full-time on template maintenance — coalescing data across spreadsheets so a single quote could be assembled.
Stale data, manual checks
Tax rates, interest rates, vehicle depreciation values — all maintained by hand. There was no easy way to know if a quote was using the latest numbers. If a tax rate changed in California, someone had to remember and update every active quote downstream.
A third party for the PDF
Once the data was finalized, the spreadsheet output was handed off to a graphic designer in Adobe to produce the polished PDF proposal. One small change to the data meant re-routing through the designer again.
Up to a full day per quote
Best case: hours. Worst case: a full day or more, especially for multi-vehicle deals or quotes that needed a small revision after the PDF was already produced. The company was capping its growth on the speed of this workflow.
John's team was facing a place where they literally could not scale the business. Quoting wasn't a slow workflow — it was the ceiling.
What we built
A custom quoting app, not a 50-tool stack.
The first deliverable was the simplest one possible: a four-step quote creation flow that absorbed all the complexity that used to live across spreadsheets and people.
A custom 4-step quote app
Replaced the spreadsheet workflow entirely. Sales pick the customer, the financing options, and the vehicles, then export. The complex inputs that used to live across templates are now built into the app's data model — depreciation schedules, term sheets, mileage options, financing tiers.
Live tax rates via state government API
Linked the customer's location to the state government's tax rate API. The right tax rate is pulled at the moment the quote is created, with manual override available. An entire class of stale-data bugs disappeared overnight.
Multi-vehicle deals with a click
Complex deals with multiple vehicles used to mean multiple template runs. The new app supports duplicating a vehicle line item and tweaking trim, term, or financing — multi-vehicle quotes now go out in the same flow as single-vehicle ones.
PDF proposals generated in-app
The graphic designer step is gone. The proposal PDF is generated automatically from the quote data, in the exact layout the company already used. Sales clicks export and the polished PDF is ready to send.
How we worked
Client-driven prototypes. Production-grade builds.
The fastest way to build the right thing is to let the people who actually do the work prototype it themselves. We set the team up to use Replit for rapid sandboxing — they captured their own processes on their own time — then walked through the prototype on Loom. We took that as the source of truth and built the production version.
- 01
Discovery and bottleneck mapping
Walked the team through their actual quoting process, sheet by sheet. Surfaced the tax-rate, depreciation, and PDF-handoff bottlenecks alongside the obvious template-wrangling problem.
- 02
Replit prototypes — driven by the client
Set the team up to prototype rapidly in Replit. They captured their own processes on their own time — the way the work actually flows when no one is watching.
- 03
Loom walkthroughs of the prototype
Once a prototype reached ~90%, the team recorded a Loom video walking through it as a real user. Captured intent, edge cases, and assumptions in their own words instead of in a spec doc.
- 04
Production build + iteration
Took the prototype intent and built the production version. Quick follow-up calls to validate, then ship. New features ship the same way: prototype → Loom → build → iterate.
The outcomes
The bottleneck moved. Then it disappeared.
Quote turnaround dropped from a full day to minutes
Best case used to be hours; worst case was a full day or more. New floor: minutes. The same sales team can now turn around a complex multi-vehicle quote in the time it used to take to update one template.
277 quotes in the first month
In the first ~30 days after launch, the team produced 277 quotes through the new system. Volume that would have been physically impossible under the old workflow.
Two data analysts freed for actual analysis
Both analysts were full-time on template wrangling. With the templates absorbed into the app's data model, they're now free to do the analysis their titles imply — pricing strategy, deal review, market trends.
Eliminated a third-party vendor
The graphic designer who produced the PDFs is no longer in the critical path. Every change downstream of a quote — a new tax rate, a tweaked term, a different vehicle — used to mean another loop through the designer. Now it's instant.
A growth path that wasn't there before
Unified Fleet is expanding into new states this year. Under the old workflow, every new state meant new tax tables, new compliance variables, and more spreadsheets to maintain. With live tax rates and a templated app, expansion is a configuration change instead of a hiring problem.
Related
Keep going.
The takeaway
The problem wasn't the spreadsheets. It was that the spreadsheets were the system.
Most small businesses don't need a 60-page AI strategy. They need someone to look at their week, find the workflow that caps their growth, and build the smallest useful thing that unblocks it. That's the work we do.