From Disconnected Systems to a Single Source of Truth
One customer, three names, and a different number in every report. I built one governed source of truth, and the arguments stopped.
The situation
A produce operation was running its business across several systems that didn’t talk to each other: a legacy produce ERP as the system of record, plus separate platforms for field data, quality, and sales reporting. Every gap between systems was bridged by a person with a spreadsheet. The same customer existed under three names. The same lot carried different numbers depending on which system you asked. Nobody was wrong. The reports disagreed because every export was pulled at a different time, cleaned a different way, and repaired by hand.
What I did
I designed and personally built a governed data warehouse on Azure SQL to sit above the operating systems as the single source of truth.
The foundation was identity: every grower, customer, item, and lot got one permanent ID, so a record means the same thing everywhere it appears. An immutable ID is just that, a name for a record that never changes even when the label on it does. Then I connected the systems through APIs, which are documented interfaces that let software exchange data directly, so information flowed automatically instead of through re-keying. The warehouse refreshes itself on a schedule, and everything downstream of it is read-only. Nobody can quietly edit a number in a spreadsheet and change the story.
Reporting moved on top of the warehouse. So did the internal tools that came later, which is the quiet payoff of doing this once and doing it right: everything built afterward inherits the same truth.
What changed
Manual re-entry between systems ended, giving back roughly 30 hours a week across the team. The three-names-for-one-customer problem disappeared, along with the reconciliation work it generated. Finance and operations now pull the same numbers from the same place, and month-end closes two days faster. When a question comes up, the answer is a query against live data, not a week of spreadsheet archaeology.