Case study
Digitized workflow transformation
Major cost and cycle-time savings plus geospatial field tools
National Grid
- Operational transformation
- Geospatial
- Field workforce
- AI / NLP
- Enterprise
- Regulated
- Workflow
Summary
Role
Principal Product Manager
Timeline
Nov 2022 – Present
Team
Operations, engineering, finance, vendors, and regional rollout teams.
Domain
Regulated utility; multi-state ops; geospatial field and supervisor workflows for service and streetlight programs.
Metrics moved
$10M+ lifecycle savings; average work close ~77 days → ~1.5 days.
Key constraints
Multi-state rules; vendor calendars; audit-ready history; offline / intermittent field connectivity.
Systems involved
Digitized workflow platform; OMW (mobile + web) with maps; work-order vendor automation; SAP Fiori; NLP on unstructured WO text; on-device capture quality.
Stack
Tech stack
OMW: geospatial supervision + crew iPad tied to work-order and finance systems—one thread from capture to systems of record.
Geospatial application layer
Supervisor web and crew iPad share maps, identifiers, and workflow state.
- PWA (crew + supervisor)
- Capacitor on iPad
- GIS-backed map views
- Pole Finder & location-aware jobs
Enterprise integrations
Field app, work order platform, and finance stay aligned.
- Enterprise work order system (bi-directional)
- SAP Fiori line items & time entry
- WO / back-office automation (vendor-supported)
Intelligence & capture
Harsh field conditions and messy payloads handled with review paths.
- NLP / structured WO text
- On-device OCR & barcode ML
- Offline-first sync & queues
How work moves through OMW
From dense system-of-record text to field-safe, reviewable execution.
Work order lands as narrative
Central platform data arrives as dense text; NLP and presentation rules make it scannable for crews and supervisors.
Crew captures truth in the field
Offline iPad flows, barcode/media capture, structured packets—IDs aligned to regulated assets.
Supervisor triage on web
Lists, maps, drawers by territory and risk—approve, reroute, drill in without losing audit or geographic context.
Systems of record stay aligned
Bi-directional WO automation + SAP Fiori line items: finance inherits the same WO and asset context the field just proved.
List & map
Supervisors triage the same jobs in list or map—drag to compare.
OnMyWay (OMW): geospatial field, iPad, and supervisor web
OMW: maps, crew iPad flows, pole/CWR capture, offline behavior, supervisor/dispatch web. PWAs from one monorepo; Capacitor on iPad where native mattered.
Screenshot 1 of 11
Overview
Principal PM for scaled digitized workflow: multi-region, vendor-heavy, executive focus on savings and days to close. Results include $10M+ savings and ~77 → ~1.5 day average close. OMW gives crews and supervisors maps, location-aware jobs, and submissions into systems finance and work management already trust—plus NLP on dense WO text and better barcode capture in the field.
Business context
Fortune 500 regulated utility: multi-state requirements, heavy vendors, zero ambiguity on accountability across internal crews and contractors.
Problem
Close cycles hit ~77 days in key territories—too slow for capital, contractors, and visibility. Paper starts and late re-entry drove errors, duplicate work, and truck rolls. Crews still re-keyed the same identifiers for finance after field capture. Unstructured WO text burned time “decoding” instead of executing.
Constraints
Jurisdiction-specific compliance; vendor and architecture limits on per-site customization; continuous delivery with defensible audit trails; offline-capable mobile without weakening asset ID discipline.
Role & ownership
Set success metrics with ops leadership; prioritized backlog by closure-time impact; aligned vendor roadmaps to internal trains; governed rollout sequencing across territories.
Goals & metrics
Lifecycle cost down, cycle time down ($10M+, 77 → 1.5 days) with compliance intact. OMW: single-source identifiers from claim through submit (worksheet, WO, Fiori aligned); structured WO text so crews and supervisors share the same picture before they leave the yard.
Approach
Mapped the true critical path to “closed” (not the org chart) and instrumented handoffs that moved dates. Piloted in a willing territory; cohort reviews to drop automation that did not touch calendar days.
Decisions & tradeoffs
Depth in closure logic over breadth of adjacent features early; territory-specific exceptions as config, not forks. Incremental cutover with parallel run where regulators needed evidence—short-term dual entry for lower rollback risk.
Cross-functional leadership
One narrative across engineering, field, finance, and vendor PMs—conflicts (e.g. sign-off order) became workflow rules and shippable UI. With CGI: automation between OMW and central WO; with SAP: field submit and “ready for time entry” use the same IDs.
Execution
Digitized workflow layer: explicit states, SLAs, integrations at systems of record. OMW: bi-directional automation with central WO (CGI removing swivel-chair steps); SAP Fiori path so worksheet submission creates line items with WO and asset billing context—end-of-day time from iPad without retyping poles/feeds/WR numbers. Crew iPad + supervisor web as PWAs from one monorepo (+ Capacitor where native mattered): offline queues, barcode/OCR where manual entry fails, NLP to normalize raw WO narrative into scannable fields. RCD and sibling programs (outage, inspection) are separate write-ups.
Outcomes
$10M+ savings and 77 → 1.5 day closes with audit trails intact. Field submissions land where finance and WM already trust; less rework and fewer follow-up visits. Structured WO presentation and capture quality cut cognitive load vs shadow spreadsheets.
What changed
Closure became a managed system: clear next owner, credible rollups, vendor incentives aligned to the same “done” definition. OMW, WO, and Fiori share identifiers by design; CGI automation replaced manual reconciliation.
Lessons learned
Fastest wins: ownership visible at handoff, not more dashboards. Shared metrics with vendors beat UI-only fixes. AI in OMW: reviewable, source-grounded output tuned on real field mess (low light, worn labels)—not demo imagery.