Semantic Layer

By defining business logic centrally, a semantic layer helps connected reports, dashboards, and AI tools interpret the same metric consistently.

A semantic layer maps business concepts to underlying data and specifies how that data should be joined, filtered, and calculated.

For example, “revenue by region” requires more than readable field names. The layer defines which revenue measure to use, how transactions relate to regions, and how amounts should be aggregated. Tools using those definitions can apply the same logic across different analyses.

• Business terms: Recognizable names and descriptions for technical fields.
• Metrics: Reusable calculations such as total sales or operating margin.
• Dimensions: Categories used to group and filter results, such as department, location, and month.
• Relationships: Defined connections between datasets.
• Governance controls: Integration with permissions and policies that determine what users can access.

These components make business meaning available to analytics tools, rather than leaving each report author to interpret raw data independently. Explore how semantic definitions work.

• Consistent reporting: Shared definitions reduce conflicting calculations across teams.
• Easier self-service: Business users can explore recognizable measures and categories.
• Simpler maintenance: Teams can manage shared logic centrally instead of editing individual reports.
• Better AI context: Defined terms and relationships help conversational analytics tools interpret business questions.

These benefits depend on adoption. Reports that bypass the shared definitions can still produce different results.

An organization defines headcount as the number of distinct employees actively employed on the final day of each month, excluding contractors. The semantic layer translates that definition into the appropriate records, date rules, and filters. An HR dashboard and a conversational query can then use the same month-end headcount measure. Without a shared definition, one report might count everyone paid during the month while another counts only employees active today. Both could be labeled “headcount” while answering different questions.

SplashBI combines enterprise data models with a governed semantic layer for reporting and analytics. SplashAI uses those models, metadata, and permissions to answer natural-language questions within the same governance framework. This supports SplashBI’s AI BI Flywheel: CONNECT enterprise data, UNLOCK its business meaning through governed models, and ASK questions that guide further analysis.