A semantic layer is the set of shared definitions that sits between raw data and the reports and tools that use it. It says what margin means, how volume is measured and which costs a cost per thousand board feet includes. Each is defined once, so every report and AI tool gives the same answer. The data tool maker dbt Labs says its version "simplifies the process of defining and using critical business metrics": "By centralizing metric definitions, data teams can ensure consistent self-service access to these metrics in downstream data tools and applications."
Why it matters
Without one, each spreadsheet and report defines measures its own way, and the same word can carry two numbers. A semantic layer settles the rule once. As the same documentation puts it, "If a metric definition changes in dbt, it's refreshed everywhere it's invoked and creates consistency across all applications."
In a timber business
The definitions that matter most are ordinary ones, such as which unit a volume is in and which costs sit inside a cost figure. Averages matter too, because they depend on what they are weighted by.
From Quarri's own work with a lumber and millwork manufacturer: blended cost per thousand board feet roughly halved while the same basket of goods rose 6%+ in price. The drop came from a change in product mix. A semantic layer would have calculated that figure the same way everywhere. It would still have fallen. A definition that states its weighting makes the cause visible. Catching it takes a second measure, such as cost for a fixed basket, defined beside the blended one.
What it isn't
A semantic layer isn't a dashboard, though dashboards read from it. It isn't the data itself. And a glossary on a wiki that no system uses doesn't count. The definitions have to be the ones the reports actually run.
Why AI makes it more important
An AI assistant asked about margin will use whatever definition it finds. In an April 2026 preprint, researchers at the semantic-layer vendor Cube gave three models 100 questions on a retail dataset. Each model answered with and without a document describing the measures. Accuracy rose by 17 to 23 percentage points with the document. Our piece on why AI needs a semantic layer covers this in more depth.
Quarri for finance and strategy teams is built for the people who close the month, explain the margin and answer the board.
Sources
- dbt Labs, "dbt Semantic Layer", documentation, last updated 18 August 2026: docs.getdbt.com
- Rumiantsau and Fokeev, "Semantic Layers for Reliable LLM-Powered Data Analytics: A Paired Benchmark of Accuracy and Hallucination Across Three Frontier Models", arXiv 2604.25149 (preprint), 28 April 2026: arxiv.org
- Quarri evidence ledger, E5 (proven)
Quarri is an AI-native data platform for the timber supply chain. It connects buying, production, sales and inventory for forest management, sawmill, wood products and pulp, paper and packaging operators.