BI Platforms · 4 min read
Looker adds six native chart types in a November visual overhaul
Looker ships six native chart types this November, retires Marketplace plugins behind them, and adds a GitHub skill that lets Claude, Codex or Antigravity write Visualization API code.
Key takeaways
- Six new native chart types ship in Looker this November: KPI visualizations, bullet charts and gauges, Sankey diagrams, hierarchical tables, histograms and network charts.
- Compact numbers auto-format $2,400,000 as $2.4M across screen sizes.
- Marketplace custom visualization plugins behind the native charts are being retired; existing dashboards continue to work.
- A new custom visualization skill in the open-source looker-skills repository on GitHub supports Antigravity, Claude and Codex.
- Filters on visualizations and Looker Data Apps remain in preview.

Looker will ship six new native chart types this November, alongside compact-number formatting, tile-level filters and an AI coding skill for custom visuals, according to a Google Cloud product post by Ani Jain, Group Product Manager at Looker, and Minhaz Kazi, in Developer Relations.
The update reaches dashboards, Explores, feeds and generative Data Apps. All of it runs on Looker's governed semantic foundation (a curated model layer that fixes metric definitions) and context from Knowledge Catalog, the catalog holding business definitions.
The post frames this as step two. Over the past year, the team rebuilt exploration around drag-and-drop workspaces, streamlined field pickers and conversational analytics, letting users start from a blank canvas. November shifts attention from how analysts explore to how the output reads in a boardroom.
What lands in dashboards this November?
Six chart types and one formatting behavior arrive first:
- KPI visualizations — core metrics with explicit period-over-period delta indicators, percentage trends and directional status.
- Bullet charts and gauges — actual performance stacked against targets, quotas and qualitative threshold bands, including a radial gauge layout.
- Compact numbers — large values auto-formatted so $2,400,000 renders as $2.4M on any screen size.
- Sankey diagrams — multi-stage conversion funnels, user drop-offs, budget allocations and material flows.
- Hierarchical tables — multi-level nested trees, such as category rollups or departmental cost centers, inside one structured tabular view.
- Histograms — frequency distributions that expose skew and outliers across numeric variables without writing table calculations.
- Network charts — entity relationships and interconnected graph nodes, aimed at surfacing clusters and bottlenecks.
A modernized dashboard follows: modern typography, refined spacing, cleaner table padding and upgraded palettes. Filters on visualizations sit in Preview, letting viewers adjust filters on individual dashboard tiles without reloading the whole page.
What does the Marketplace retirement mean?\n
Several of these charts began as custom visualization plugins in the Looker Marketplace. Google Cloud now supports them natively and says it is retiring the associated plugins from the Marketplace. Existing dashboards using those plugins continue to work, per the post. The announcement does not name the affected plugins, sets no removal date and describes no migration path beyond continuity. Teams that depend on them should inventory their instances and record the underlying LookML definitions now.
How much JavaScript does the analyst still write?\n
Previously, building a custom chart against Looker's Visualization API (the interface that renders third-party visuals inside Looker) meant writing JavaScript yourself or waiting for a front-end developer. A new custom visualization skill in the open-source looker-skills repository on GitHub changes that equation. It works with AI coding agents including Antigravity, Claude and Codex. The analyst describes the chart in natural language, the agent writes code against the Visualization API, the design is refined conversationally, and the agent deploys it.
"Nobody on your team needs to know JavaScript," Jain and Kazi wrote.
The post names the libraries in play: D3, Vega-Lite, ECharts, Chart.js and React, or raw SVG (a text-based vector format) drawn from scratch. The finished chart pulls data from a LookML model (Looker's modeling layer) and inherits native drill menus, cross-filtering and dashboard controls. Generated examples include sunburst, KPI radar and calendar heatmap charts.
The same visual system and agent-driven workflow powers Looker Data Apps, still in preview: an agent assembles a multi-step application sharing the modernized typography, palettes and governed LookML metrics.
What is missing from the announcement?\n
The post gives no version number, no date beyond "this November", no pricing detail and no benchmark for agent-written chart quality. It offers no measure of hours saved and no code-review policy for output an agent produced. The governance claim carries no independent check: "Whether you're adding a bespoke chart to an existing dashboard or building a standalone application, every view stays consistent and grounded in the same semantic model," the authors write.
What should a data team do next?\n
- Try the native types in a Looker environment today; the post points to the Looker documentation for details.
- Run the custom visualization skill in your own Looker project through the open-source looker-skills repository on GitHub, and read the generated code before it reaches production.
- Benchmark agent output against hand-built charts on your own metrics before standardizing on it.
- Keep a named JavaScript owner for charts under audit or regulatory scrutiny, since the post documents no review process.
- Inventory dashboards built on the retiring Marketplace plugins and capture their LookML definitions before removal.
Google News: Business Intelligence (Source)
More from Tom Whitfield
Show full bio
Staff writer covering regulation, policy and their impact on Analytics Digest is a professional daily at Analytics Digest.
16 articles