Real-time analytics for Nimbus Logistics
From overnight Excel reports to real-time dashboards that drive same-day dispatch decisions across 14 distribution centers.
- Client
- Nimbus Logistics
- Published
- January 2026
The setup
Nimbus runs same-day distribution across 14 centers. Their dispatch team was making routing calls against data that was, at best, six hours stale — pulled from an overnight snapshot of the WMS into a labyrinth of linked Excel workbooks.
The cost was hidden but consistent: dispatch couldn't trust the inventory counts, so they over-allocated to be safe. Over-allocation meant inflated trucks, more empty miles, and roughly 7% margin compression they'd accepted as the cost of doing business.
What we built
A real-time data layer on top of their WMS and TMS, plus a custom dashboard surface for the four roles that actually drive operations: dispatcher, regional ops manager, hub lead, and exec.
What changed
Before
- WMS → Excel via overnight CSV
- 12-minute median dispatch latency
- 7% margin compression from over-allocation
- 4 analysts owned all reporting
After
- WMS → Snowflake via Fivetran, < 90s freshness
- Real-time dashboard per role
- Margin recovered to baseline within Q3
- Self-serve metric layer (dbt + Metabase)
We stopped scheduling dispatch around the data freshness lag. The team makes calls in real time now.
Where they are now
The dashboards are running in production at all 14 hubs. Daily query spend dropped from $340 to $18 after we tuned the materialization schedule and pushed hot reads to result-caching. The analytics team scaled from 4 to 6, but they're now building new models instead of reconciling old ones.
We stopped scheduling dispatch around the data freshness lag. The team makes calls in real time now.