Looker

What it is

Looker is Google Cloud’s enterprise BI platform, built around LookML — a code-defined semantic layer where your data team writes, once, what “pipeline,” “net revenue retention” or “active account” means. Every dashboard, scheduled report and AI answer then queries your warehouse live through those definitions. It runs against BigQuery, Snowflake, Databricks, Redshift and most other SQL databases, and ships in two forms: Looker (original) and Looker (Google Cloud core), the version billed through a Google Cloud account. Google named it a Leader in Gartner’s 2026 Magic Quadrant for Analytics and BI Platforms, its third year running.

The 2026 story is agents. At Google Cloud Next on April 22, 2026, Google announced Looker BI Agents: the Conversational Analytics agent (already GA) got a reasoning upgrade in preview, dashboard agents answer questions inside an existing dashboard, and scheduled agentic workflows that watch metrics for anomalies entered preview on October 9. Data agents for both user-defined and LookML dashboards became generally available on October 8. A Looker-managed MCP server is in preview, alongside open-source MCP Toolbox support, and the Looker extension for VS Code went GA on September 10.

Not to be confused with Data Studio: Google renamed Looker Studio back to Data Studio in April 2026, so “Looker” now means only the semantic-layer platform.

Why it shows up in RevOps and CS stacks

• One metric definition across teams. The recurring RevOps fight is finance, sales and CS each reporting a different ARR. In Looker the definition lives in version-controlled LookML, and every chart, schedule and agent answer inherits it. That is the strongest reason to choose Looker over a spreadsheet-style BI tool.
• Agents answer from governed definitions, not raw tables. A CSM asking “which accounts dropped usage 30% this quarter?” gets an answer computed from the same LookML measure the QBR dashboard uses. Data teams can mark verified queries (GA since August 28) so common questions return a vetted answer.
• It reaches Gemini Enterprise. Conversational Analytics agents can be published into Gemini Enterprise, and since August 5 those chats return Looker charts. Engineers can query the semantic layer from Claude or Cursor through MCP.

Pricing reality

Google does not publish annual prices. Its pricing page lists three platform editions — Standard (under 50 users, 1,000 query-based API calls a month), Enterprise (100,000) and Embed (500,000) — each including 10 Standard and 2 Developer users, sold on one-, two- or three-year terms, plus per-user licenses in three types: Developer, Standard and Viewer. Third-party readings of Google’s SKU list, checked October 12, 2026, put Standard at about $5,000 a month, with extra users at roughly $30 Viewer, $60 Standard and $125 Developer per month.

Buyer data: Vendr reports a median Looker contract of $84,602 a year across 418 purchases, with buyers saving about 18% off the first quote. Budget $60,000 a year as the floor for Standard before extra seats, and $85,000–$150,000 for Enterprise with a few hundred viewers. Warehouse compute is billed separately.

Conversational Analytics is metered in “data tokens.” Standard includes 60M input and 1.2M output tokens a month, Enterprise 300M and 6M, Embed 1.2B and 24M. Overage is $3 per 1M input and $20 per 1M output tokens, but billing is not yet enforced: Google is running a promotional no-quota period and has committed to 90 days’ notice before charging.

Best for

RevOps and CS ops leaders at companies with 200 or more employees, a modeled warehouse (BigQuery especially) and at least one analytics engineer who will own LookML, where the main problem is conflicting metric definitions across 50+ report consumers — and where those consumers increasingly ask questions in chat rather than open dashboards.

Skip it if nobody on staff will write LookML; you will pay enterprise prices for a tool that stalls at its first model. Skip it if you are a Microsoft 365 E5 shop, where Power BI Pro is already in the bundle.

Versus the alternatives

Power BI leads BI by installed base at $14 per user per month for Pro. Pick it for a Microsoft shop where reporting is the whole job. Pick Looker when you need a code-reviewed semantic layer shared across tools and agents.

Tableau Cloud Creator lists at $75 per user per month. Pick it for trained analysts doing heavy visual exploration. Pick Looker when governance of shared definitions matters more than visual freedom.

Omni is the fastest-growing entrant — founded by former Looker executives, it raised a $120 million Series C at a $1.5 billion valuation in April 2026 on 4x year-over-year revenue growth. Pick Omni if you want a Looker-style semantic layer with faster spreadsheet-style self-serve and are not committed to Google Cloud. Pick Looker for BigQuery depth and Gemini Enterprise reach.

Sigma wins when business users must write back to the warehouse. Hex wins when the work is exploratory analysis in SQL and Python notebooks. If none fit because your source data is not in a warehouse yet, start with data warehouse vs CDP.

Watch-outs

• LookML is a developer job. Self-service Explores and the Expression Assistant help, but every durable metric still lives in LookML. Guard: name one analytics engineer as LookML owner before signing, and budget Developer seats for them; treat self-service Explores as prototypes that get promoted into LookML.
• Conversational Analytics overage will start billing. The free period ends on 90 days’ notice, and Google has not published how many tokens a typical question uses. Guard: enable the token-usage observability preview now, baseline 60 days of use, and write the included allowance and overage rate into your renewal.
• Gemini Enterprise integration is still settling. Published data agents must be re-published after a Gemini Enterprise instance update. Guard: pilot with one team, and add “re-publish Looker agents” to your Gemini Enterprise change checklist.
• 2026 deprecations remove things you may use. Looker reports were deprecated on July 13, 2026; the standalone Spectacles service shuts down from November 30, 2026; complimentary Data Studio Pro licenses ended for Google Cloud core instances created after August 1, 2026. Guard: move report-style output to paginated reporting (GA), move Spectacles checks into Looker Continuous Integration before November 30, and price Data Studio Pro separately in new deals.
• Looker queries databases, not apps. It has no native CRM or support-desk connectors. Guard: land Salesforce and Zendesk data in the warehouse first with Fivetran or similar, and include that ELT cost in the business case.