Generate Looker LookML View Files

Creates LookML view files for Looker - covering dimensional modeling, measure calculations, derived tables, and SQL patterns structured for query performance and business-user usability.

Use this skill when writing a new Looker view from a database table, adding measures or dimensions to an existing view, building a derived table for cohort or other complex logic, or optimizing a slow-performing view. Not a fit for building Looker Explores/dashboards themselves (this generates view-level LookML, not dashboard layout) or for BI tools other than Looker. The guidance is framed around end-user experience and query performance as the two considerations that should drive every view design decision.

Every view follows a fixed structure: connection and table declaration, a primary key dimension, dimensions, measures, derived tables where needed, and field documentation. A basic view template shows sql_table_name, a primary key dimension, a dimension_group for timestamps with multiple timeframes, a string dimension with a drill-through link to a dashboard, count and sum measures, and a set grouping fields for drill fields.

Dimension patterns include dimension_group for time fields with explicit timeframes and timezone handling, categorical dimensions using case blocks to map raw codes to business labels, and geographic dimensions using map_layer_name and sql_latitude/sql_longitude for location types.

Measure patterns include conditional measures with filters, ratio measures like conversion rate using NULLIF to guard against division by zero, running_total and percent_of_total measure types for window-function-style aggregation, and a derived table example computing user cohorts by first-order month with sql_trigger_value for incremental rebuilds.

Performance guidance covers always defining primary keys for join performance, schema-qualifying sql_table_name, efficient yesno filter dimensions for quick filtering, and a default date-range filter. A data quality measure example combines a percent-formatted score with conditional HTML/Liquid coloring (green/orange/red) based on the value.

Generates LookML for the Looker BI platform, targeting SQL-based warehouse tables and using Looker-specific constructs (dimension_group, derived_table, sql_trigger_value, Liquid templating in html fields).

Use descriptive names matching business terminology, document every field for business users, define logical drill paths, apply appropriate value formats (currency, percent, decimal places), use efficient SQL patterns while avoiding unnecessary calculations, validate measures against known results before deployment, apply appropriate caching strategies for derived tables, and implement access grants for sensitive data.

Analytics engineers and BI developers building or maintaining Looker views who need correct LookML structure, measure patterns, and derived-table logic rather than a general SQL or dashboard-design guide.

measure: conversion_rate {
  type: number
  sql: 1.0 * ${converted_users} / NULLIF(${total_users}, 0) ;;
  value_format_name: percent_2
}