At $8B in annual spend and 800,000+ contracts a year, contract scoping actually lived in Excel files emailed back and forth — and Scopeworker had quietly become a transcription layer, adding 1–2 weeks before work even began. I re-architected contract creation so suppliers author scope from on-site reality, while buyers keep approval and control.
The strategic bet was to separate authorship from authority: let the people with ground truth — suppliers on site — write the scope, while the paying customer keeps approval, compliance, and downstream control. That single move aligned the system to how work actually happens without shifting any power, and let the platform replace Excel as the system of record instead of competing with it.
Scopeworker is the enterprise ERP US telecom operators like T-Mobile and Samsung use to run field execution for network build and maintenance. A contract is the core execution unit: it defines what work happens at a site, how it's done, and what the supplier gets paid, then rolls up into POs, invoices, and close-out packages.
At the time of this project, one thing was throttling an otherwise scalable platform: creating those contracts. Over 800,000 are created a year — north of $8B in spend. The prime client was T-Mobile, the largest US operator.
By design, contract creation was buyer-led: buyers defined the scope of work in the platform and issued contracts to suppliers. In reality, that's not how the work happened.
Suppliers were the ones visiting sites, assessing on-ground conditions, and determining what actually needed doing — and none of that was captured directly in Scopeworker. Suppliers shared findings in Excel files over email, which buyers then manually re-typed into contracts.
The system of record was an Excel file in someone's inbox. The platform wasn't where scoping happened — it was where scoping got re-typed after the fact. Every contract started with delay and information loss baked in.
Fixing it meant touching who does what — which is exactly the kind of change that fails on adoption if you get it wrong.
I evaluated three approaches. Two of them treated symptoms:
Option C is the only one that aligns the system with reality without moving power. Here's the shift, made explicit:
Only one row changes — who authors. Authority, approval, and compliance stay exactly where they were. That's the whole reason it was adoptable.
An in-platform, table-based authoring experience that felt like the tools teams already used — so structure and familiarity survived the move into a governed system. It also carried context Excel never could, like structured line-item information.
| Line item | Qty | Unit | Rate | Amount |
|---|---|---|---|---|
| Foundation pour — tower base | 1 | ea | $8,200 | $8,200 |
| 40ft crane rental | 2 | day | $1,450 | $2,900 |
| Antenna install crew — 3 sector | 1 | crew | $3,400 | $3,400 |
| Fiber pull — 400m | 400 | m | $6.50 | $2,600 |
Multiple users across both organizations could edit one scope, with every change tracked, attributable, and auditable — and an email notification whenever a new version was published. Buyers kept final approval authority throughout.
The subtle part: I decoupled who creates the data from whose rules apply. Buyer validations kept running against supplier-authored scopes, so compliance and downstream-system compatibility held even though the author had changed.
The goal was never feature completeness — it was making Scopeworker the real system of record, fast, without a risky big-bang.
Contract creation cycles dropped sharply, buyers gained earlier visibility and clearer control, and suppliers became first-class participants instead of email attachments. Scopeworker went from recording contracts after the fact to being the place they're actually created.
Good enterprise products don't enforce the ideal process — they adapt to how work actually happens, while preserving control, accountability, and trust. The win here wasn't a slicker form. It was designing the system around reality instead of the org chart.
The adoption number is hard and clear (~60%). The cycle-time win I can describe but not yet quantify to the day — the version of this story with a precise before/after on scoping cycle time is the stronger one, and that's the first thing I'd instrument on a re-run.