A single source of truth means each fact (a contract value, a cost code budget, an RFI status) lives in exactly one place, and everything else reads from it. A data core is the shared database and data model that makes that possible across project management, accounting and reporting. The opposite is the usual reality: the same change order typed into a PM tool, an ERP, a spreadsheet and a BI model, each slightly different.
Why it matters
When numbers disagree, people stop trusting reports and start building their own spreadsheets, which makes it worse. A single source of truth removes reconciliation work and shortens the time between something happening on site and the forecast reflecting it.
Worked example
Illustrative: on Riverside Medical Center, the PM’s report shows contract value of $49.3M ($48.2M plus $1.1M approved COs), but the ERP shows $49.21M because CO #14 ($86,400) and a smaller CO weren’t entered yet. Neither is wrong on its own terms. With one record, there is only one contract value, and it changes the moment the CO is approved.
Common mistakes
- Calling a data warehouse that syncs nightly a “single source of truth.”
- Letting teams keep shadow spreadsheets because the system is hard to use.
- Not deciding which system owns each field when multiple tools must coexist.
How os.construction handles it
A data core is the foundation of what we’re building: project management, accounting, analytics and agents reading and writing one model. We’re building it with founding contractors.