Replacing Manual KPI Reports with a Real Data Platform
How we rebuilt a Canadian company's KPI reporting on a Postgres-to-Postgres data platform, engaged as an external contractor.
Challenge
The company's KPI reporting ran on legacy dashboards that were slow to update and hard to extend. Reports covered two separate deal streams, and every new metric or breakdown meant more manual, hand-built work rather than a change to a shared data model. There was no single place where the CRM, marketing, and advertising data actually lined up.
Approach
We built a Postgres-to-Postgres reporting architecture: source data landed in Postgres, and dbt modeled it through staging, intermediate, and mart layers so each deal stream had its own clean, documented set of metrics built on shared logic rather than duplicated queries.
For the dashboard layer, we evaluated Power BI, Looker, and Lightdash alongside Preset (hosted Apache Superset), and selected Preset on cost and scale for the company's size. Dashboards drew on the CRM, Zoho, Google Ads, and Meta Ads, giving the team one place to see pipeline, marketing spend, and ad performance together.
We ran the build through a series of stakeholder feedback sessions, refining metric definitions and dashboard layouts against how the team actually used the reports day to day.
Result
- Legacy KPI reports across both deal streams replaced by a single set of Preset dashboards on shared dbt models
- Manual report-building gave way to dashboards that update automatically as new data lands
- Metric definitions and dashboard layouts refined directly with stakeholders, matching how the team actually tracked performance