How many hours per month do we spend reconciling numbers that should match but don't?

Reconciliation time is a direct measure of semantic inconsistency. Every hour spent figuring out why two reports differ is an hour that would not exist if both used the same definitions.

If we defined every key metric once and enforced it everywhere, what problems would disappear?

Metric conflicts, report reconciliation, dashboard distrust, and meetings about the numbers instead of the decisions all disappear when definitions are codified and enforced.

Data & Analytics Intelligence

Semantic layer

Q: When our CFO says 'revenue,' do they mean billed, collected, or projected?

If different people mean different things by the same word, every report is a potential misunderstanding. A semantic layer codifies these definitions once so every system uses the exact same meaning.

By Mark Ziler · Last updated 2026-04-05

When your CFO says 'revenue,' do they mean billed, collected, or projected? When your COO says 'productivity,' is that total hours or billable hours? A semantic layer answers these questions once so that every report, every dashboard, and every AI agent uses the exact same definition. Without it, two people pull the same metric and get different numbers — and the meeting becomes about reconciling spreadsheets instead of making decisions. The semantic layer isn't glamorous technology. It's the agreement on what your numbers mean, codified so machines enforce it consistently.

Go deeper

Your new CFO starts Monday. She pulls the revenue report and sees $4.2M for last quarter. She pulls the ops dashboard and sees $3.9M. She asks your controller, who says it is actually $4.1M. On her second day, she has to explain to the board why nobody in the company agrees on revenue. The problem is not the data — it is that each report uses a different filter, a different date range logic, and a different treatment of credits and adjustments. A semantic layer would have made all three numbers identical because the definition of 'revenue' would exist in exactly one place.

The trap most companies fall into is letting the semantic layer become a technical artifact that only the data team understands. The definitions need to be co-authored with finance, operations, and clinical leadership — the people who will be held accountable for the numbers. If the data team defines 'productivity' without input from clinical ops, they will get it technically correct and operationally meaningless. The semantic layer is a business document that happens to be machine-readable.

Questions to ask

If I asked five leaders in our company to define 'revenue,' would I get five different answers?
Who was involved in defining our current KPI calculations — and did operations, finance, and clinical all sign off?
When was the last time someone audited whether the numbers on the dashboard match the definitions in the data dictionary?