LookML semantic layer
LookML is the reason to buy Looker and the reason it's so expensive — a proprietary modelling language that needs a dedicated analytics engineer to maintain.
What people actually say
✕Vendor cost analyses estimate organisations spend 40-60% of their total Looker investment on LookML development and maintenance, with initial model build running $20,000-$100,000 and ongoing maintenance at 0.5-2 FTE.
Source: Mammoth, Looker Pricing 2026 ↗✕LookML is a proprietary modelling language with a steep learning curve that adds implementation costs on top of the licence — pricing guides treat 'specialised LookML engineers' as a baseline staffing assumption, not an optional add-on.
Source: Holistics, Looker Pricing 2026 ↗✕Industry repositioning frames Looker as a high-cost semantic layer where the modern split is dbt building the marts and LookML only describing how analysts explore them — i.e. you're paying for one layer of the stack you can now do in your repo.
Source: Valiotti Data, Looker Two Years After Google ↗
The semantic layer lives in dbt as version-controlled SQL with tests, exposures and a generated docs site — the explicit open replacement for LookML that the industry has standardised on. Metrics are pull requests reviewed in GitHub, not artefacts inside a Looker instance you rent annually. Migrating off later is `dbt run`, not a re-platform.