01The Problem
What wasn't working
A mid-market logistics company had data scattered across five systems and no way to answer basic questions about fleet performance.
02The Challenge
Where it started
The five source systems had no shared identifiers: the same vehicle appeared as different codes in the fuel system, the ERP, and the dispatch tool. Cleaning and joining this data was the entire problem — the dashboard was the easy part.
03Research & Planning
Finding the real problem
We sat with the finance and operations teams and listed the 40 questions they asked most often about the fleet. These became the metric definitions. We then audited each source system to map fields, formats, and refresh cadences.
04Design
Designing for the people who use it
We used a KPI-first layout: every dashboard opens with the few numbers that matter, then lets users drill into detail. Color coding follows a strict convention (green = good, amber = watch, red = action) so the dashboards read correctly at a glance.
05Architecture
The technical blueprint
Nightly batches load raw data into ClickHouse for fast aggregations, and dbt models transform it into a clean metrics layer. The Next.js front end queries pre-aggregated tables so dashboards stay fast even over slow office connections. An alerting service watches fuel and utilization patterns and notifies managers via email and chat.
06Development
Building in shippable increments
We started with the finance dashboard because cost questions were the highest priority, then added operations and executive views. Data validation checks ran against the client's existing monthly reports to prove correctness before each module went live.
07Final Solution
What shipped
A cloud analytics platform that ingests, models, and visualizes operational data in one place — with role-based dashboards.
ArchData now answers fleet questions in seconds instead of weeks. The monthly reporting process that consumed three people-days is now a scheduled task. The data team, previously stuck reconciling spreadsheets, now spends its time investigating the anomalies the platform surfaces.
08Key Features
What the product does
- Automated ingestion from fleet, ERP, and fuel systems
- Data warehouse modeled with dbt for consistent metrics
- Executive, operations, and finance dashboard suites
- Drill-down from company-level KPI to individual vehicle
- Scheduled report delivery via email
- Anomaly alerts on fuel and utilization patterns
09Technology
The stack
- Next.js
- TypeScript
- PostgreSQL
- ClickHouse
- Python
- dbt
- AWS
