Visualisation & BI Plane · BI & Dashboards

Looker

BI platform on Google Cloud with a modelling layer, LookML, for governed metrics.

Overview

Looker is the only tool in this category that requires a separate modelling layer. Its modelling language is mandatory, version-controlled, and there is no import or extract mode to fall back on.

That has a direct consequence for cost and architecture: it always queries the warehouse live, so your warehouse bill is your analytics bill, and materialisation is handled by persistent derived tables rather than an in-memory engine.

It is a Google Cloud service rather than a standalone product, and every instance has a cloud billing account attached, which is the clearest lock-in among the tools here.

Two flavours still coexist in its own documentation, the original product and the cloud-core one, with release notes calling out features that land on only one of them.

Do not confuse it with the free self-serve product of a similar name, which is a separate tool with separate pricing, and whose older name still appears in places in the navigation.

Its 2026 direction is agentic: conversational agents attach to explores and dashboards, with a debug mode, verified query pairs as ground truth, and per-agent observability.

A managed protocol server now lets external assistants connect without middleware, and usage is tracked in the product, which is a more governed answer than most.

A genuine shift is that the modelling layer is no longer closed: dimensions and measures can be exported into in-database semantic models in two major warehouses.

Two dated items matter for planning: its report surface was deprecated in July 2026, and the legacy mobile application retires at the end of January 2027.

Its commercial gap is significant for buyers. Platform and per-user prices are sales-gated, and the only published figures are the assistant's data token rates.

Key features and capabilities

The same headings are used for every bi & dashboards entry, so two tools can be read side by side.

Data modelling and semantic layer
  • A mandatory modelling language holding dimensions, measures, explores and joins in version-controlled files
  • There is no import or extract mode; it always queries the warehouse live
  • Persistent derived tables and aggregate awareness handle materialisation and performance
  • Modelled dimensions and measures can now be exported into in-database semantic models in two major warehouses
  • That export is a real change from a closed semantic layer to a portable one
Charts, dashboards and interactivity
  • Single charts, dashboards and boards, with a built-in library plus marketplace visualisations
  • Cross-filtering on dashboards, and drill to row-level detail as a granted privilege
  • Custom visualisations install from its marketplace
  • Dashboard agents now appear inside embedded dashboards for permitted embed users
  • The legacy mobile application retires at the end of January 2027
Calculations, AI and natural language
  • Measures defined in the modelling language, plus raw SQL where needed
  • Conversational agents with a thinking mode, verified query pairs as ground truth and a five-minute query ceiling
  • Key-driver analysis can now be attached to alert notifications
  • Scheduled alerts on content
  • Agents publish into the wider enterprise assistant product
Publishing, embedding and collaboration
  • Scheduled delivery of looks and dashboards
  • Embedding through signed or federated sign-on plus an embedding toolkit
  • Actions push results into external systems
  • Export to delimited, spreadsheet, structured and image formats
  • Its own report surface was deprecated in July 2026
Security, lineage and certification
  • Row and column-level access through access filters, grants and user attributes in the model
  • Version control is native to model development, with a development mode and pull requests
  • Content validation checks that changes do not break existing content
  • System activity dashboards for usage, now including assistant token observability
  • Administrators can transfer ownership of dashboards, boards and agents programmatically
Integrations
  • Connects to essentially all major SQL warehouses
  • Its published interface carries per-edition quotas on query and administrative calls
  • A marketplace supplies blocks, applications and visualisations
  • A managed protocol server lets external assistants connect without middleware
  • European data residency now applies to all conversational agents
Where it runs and what it costs
  • Hosted by Google, or customer-hosted for the original flavour
  • Locked to a Google Cloud billing account, which is the clearest tie among these tools
  • Every platform edition includes one production instance and a small number of seats
  • Pricing is a platform subscription plus named-user licences plus assistant data tokens
  • Regions and sovereign options were not confirmed in this pass

Pricing

Price on requestQuote only; token rates published

Sales-gated in the places that matter. Every platform edition on the pricing page says to call sales, and per-user rates for its three licence types are likewise unpublished, which makes this the least transparent tool in the category. The only published figures are for assistant usage: three dollars per million input data tokens and twenty per million output tokens as overage, with monthly allowances pooled per instance that scale by edition and do not roll over.

Vendor pricing page →

Demos and videos

About Google

Looker is operated by Google Cloud, part of Alphabet, and is not a standalone vendor: instances are tied to a cloud billing account. Alphabet reported second-quarter 2026 revenue of $119.8bn with cloud revenue growing eighty-two per cent and a cloud backlog of $514bn, though none of those figures is specific to Looker, and Looker's own revenue, headcount and customer count are not published. The acquisition dates from 2020 but its terms were not verified from a first-party page in this pass. One naming inconsistency to note when researching: the platform tables on its pricing page name one set of editions while the token allowance table names two others, and the extra names are not defined anywhere else.

cloud.google.com

Other bi & dashboards tools

Apache Superset

Visualisation & BI Plane · BI & Dashboards

Open-source data exploration and visualisation platform.

  • Open source

Domo

Visualisation & BI Plane · BI & Dashboards

Cloud analytics platform spanning connectors, transformation, dashboards, workflow automation and AI, priced by consumption credits with unlimited users rather than per seat.

  • Commercial

Lightdash

Visualisation & BI Plane · BI & Dashboards

BI built natively on dbt, taking metrics from the dbt project as the semantic layer. Open source to self-host, with a cloud tier priced per month rather than per seat.

  • Open core

Metabase

Visualisation & BI Plane · BI & Dashboards

Open-source BI tool for questions, dashboards and embedded analytics.

  • Open core

Omni

Visualisation & BI Plane · BI & Dashboards

BI that pairs a shared data model with raw SQL, building the model progressively out of the analysis people actually do rather than requiring it up front.

  • Commercial

Power BI

Visualisation & BI Plane · BI & Dashboards

Microsoft's BI platform for semantic models, reports and dashboards, integrated with Fabric.

  • Commercial

Qlik Sense

Visualisation & BI Plane · BI & Dashboards

Analytics platform with an associative engine for exploring data.

  • Commercial

Sigma

Visualisation & BI Plane · BI & Dashboards

Spreadsheet-style analytics that queries the cloud warehouse directly, covering ad hoc analysis, dashboards, data apps and embedded analytics without extracting data.

  • Commercial

Tableau

Visualisation & BI Plane · BI & Dashboards

Visual analytics and BI platform for building interactive dashboards.

  • Commercial

ThoughtSpot

Visualisation & BI Plane · BI & Dashboards

Search and AI-driven analytics on cloud data, sold as an analytics product and an embedded SDK, with LLM usage included in the subscription rather than metered.

  • Commercial

Drafted with AI assistance and checked against the vendor’s own documentation.