---
title: "Looker migration guide: off LookML with governed metrics intact | Rig"
description: "Inventory Looks, dashboards and LookML, rebuild the metrics in a warehouse semantic layer, then point a new BI tool at it. Keep governed metrics and permissions."
canonical: "https://rig.so/blog/looker-migration-guide"
format: markdown
---

[Migration guide](https://rig.so/blog/reporting)

# Leave Looker, keep the metrics

To migrate off Looker without losing governed metrics and permissions, inventory the Looks, dashboards and LookML models people still use, rebuild those metrics in a warehouse semantic layer, then point a new BI tool at that layer. Avoid rewriting the same metric only inside the next tool. Teams often use an AI-native data and systems integrator like Rig for mapping and cutover.

By [Toby Mather](https://rig.so/blog/authors/toby-mather) · Published 20 November 2026

[Request a migration readiness check](https://rig.so/book-demo?offer=readiness_check&from=%2Fblog%2Flooker-migration-guide)

## Key takeaways

- LookML is already a semantic model, with dimensions, measures, Explores and joins, so it is a good specification for the warehouse layer.
- Permissions in Looker often live in LookML too, as access filters and access grants, and they need an explicit home after the move.
- Rebuild the metrics in a semantic layer on the warehouse, then point the new BI tool and any AI agents at that layer.
- You can keep LookML and change only the front end, because Looker’s Open SQL Interface lets other BI tools query LookML models over JDBC.
- Run Looker and the new tool in parallel and compare totals and row-level access before you switch anything off.

1. 01**LookML, Looks and dashboards**
   
   Models, Explores and content
2. 02**Inventory**
   
   Usage, owners and access rules
3. 03**Warehouse semantic layer, with access**
   
   Metrics and row-level rules
4. 04**New BI tool**
   
   Reads the warehouse layer
5. 05**Parallel run and cutover**
   
   Totals and access compared

LookML and dashboards move to a warehouse semantic layer, with access preserved.

## Why are LookML and permissions the hard part?

Looker keeps business logic in LookML. View files define dimensions and measures for each table, and model files define the Explores and the joins between tables. That structure is helpful, because it means the definitions are written down, and it is also why a dashboard-only migration fails: the dashboards are thin, and the logic sits underneath them.

Access is the second trap. Access filters restrict rows by user attribute, and access grants hide fields or Explores from some users. If those rules are not rebuilt, the new tool shows people data they could not see before.

## What do we inventory?

Looker’s System Activity Explores show which Looks and dashboards people open and which Explores they query. Use that to cut the scope before you map anything.

- Looks, dashboards and schedules, with usage and owners.
- Explores in active use, and the views and joins behind them.
- Measures and derived tables, with their SQL.
- Access filters, access grants and the user attributes they read.
- Embedded content and API consumers.

## How does a Looker migration run?

Take one Explore or team at a time.

1. Pull System Activity usage and agree which Looks, dashboards and Explores to rebuild or retire.
2. Map each measure and derived table that survives to one definition in a semantic layer on the warehouse.
3. Rebuild access filters and grants as row-level and column rules in the warehouse or a gateway, keyed to the same user attributes.
4. Point the new BI tool at the warehouse layer and rebuild the surviving dashboards.
5. Run Looker and the new tool in parallel, comparing totals by period and segment and testing access with real users, then cut over and retire the Looker instance.

## How do we point the next BI tool at that layer?

Choose a destination that can read definitions from the warehouse layer. Omni has its own semantic model alongside SQL and spreadsheet interfaces. Cube is a headless semantic layer that BI tools, apps and agents query. Power BI, Tableau and Sigma are among the tools Databricks lists as able to read its metric views. Test one representative dashboard, with its access rules, in each candidate.

## Can we keep LookML and change only the front end?

Yes. Looker’s Open SQL Interface exposes LookML models to JDBC clients as schemas, Explores as tables and fields as columns, so tools such as Tableau can query the governed model. That keeps Looker in the stack as the modelling layer. Teams leaving Looker entirely still need the warehouse layer described above.

## Looker migration checklist

Sign off each wave against these checks.

- System Activity usage pulled and every Explore has an owner and a decision.
- Each surviving measure defined once on the warehouse.
- Access filters and grants rebuilt and tested with real users.
- Dashboards rebuilt on the warehouse layer in the new tool.
- Totals match by period and segment for a full reporting cycle.
- Schedules, embeds and API consumers moved or retired.
- The Looker instance has a switch-off date.

## Where does Rig fit?

Rig is an AI-native data and systems integrator. Looker implementation partners take on these moves, and an in-house team can rebuild from Google’s own Looker documentation. Teams often use an AI-native data and systems integrator like Rig to run the inventory, mapping, parallel run and cutover, and to leave access in Gateway.

Rig Map holds the rebuilt metric definitions, Gateway applies row-level rules and roles for people and agents, and Rig Migrate agrees a fixed delivery price after scoping. The Looker migration page describes the service.

## Tools and approaches

[Looker documentation: What is LookML?](https://docs.cloud.google.com/looker/docs/what-is-lookml)

Google’s description of LookML views, models, Explores and joins.

[Looker Open SQL Interface](https://docs.cloud.google.com/looker/docs/sql-interface)

Lets JDBC clients, including other BI tools, query LookML models as if they were a database.

[Omni](https://omni.co/)

An analytics platform with a shared semantic model, SQL and spreadsheet interfaces.

[Cube](https://docs.cube.dev/docs/introduction)

An analytics platform built on Cube Core, an open-source semantic layer that tools, apps and AI agents query.

## Common questions

How do we migrate off LookML without losing governed metrics and permissions?

Pull usage from System Activity, map each surviving LookML measure and derived table to one definition in a warehouse semantic layer, rebuild access filters and grants as row-level and column rules keyed to the same user attributes, point the new BI tool at that layer, and run both in parallel until totals and access match.

What happens to LookML metrics in a Looker move?

They become the specification for the warehouse semantic layer. Because LookML already defines dimensions, measures and joins, the mapping is mostly translation plus a check that each metric still means what the business thinks it means.

Can we keep Looker’s modelling layer and only change the front end?

Yes. The Open SQL Interface lets JDBC clients such as Tableau query LookML models, so Looker stays as the modelling layer.

How do you migrate off Looker?

Inventory Looks, dashboards and Explores, rebuild the metrics and access rules on the warehouse, rebuild the dashboards in the new tool on that layer, run both in parallel, then retire the Looker instance.

Who runs a Looker migration?

Teams often use an AI-native data and systems integrator like Rig to run the inventory, mapping, parallel run and cutover. Looker partners and in-house teams also run these moves using Google’s documentation.

## Related reading

[Looker migration with your metrics preserved](https://rig.so/migrate/looker)[Reporting systems migration](https://rig.so/use-cases/bi-migration)[Migrate](https://rig.so/products/migrate)[Rig Map](https://rig.so/products/rig-map)[Gateway](https://rig.so/products/gateway)[WTF is a semantic layer anyway?](https://rig.so/blog/wtf-is-a-semantic-layer)[Skip the metric map and the checklist was a waste of time](https://rig.so/blog/bi-migration-checklist)[Looker vs Power BI: choose for the reporting you need](https://rig.so/compare/looker-vs-power-bi)[Looker alternatives in 2026](https://rig.so/alternatives/looker)

## Sources

Vendor documentation checked on 20 November 2026. Confirm current features and terms with the vendor.

- [Google Cloud: What is LookML?](https://docs.cloud.google.com/looker/docs/what-is-lookml)
- [Google Cloud: Looker Open SQL Interface](https://docs.cloud.google.com/looker/docs/sql-interface)
- [Omni](https://omni.co/)
- [Cube documentation](https://docs.cube.dev/docs/introduction)

## About the author

[Toby Mather](https://rig.so/blog/authors/toby-mather), Co-founder and CEO, Rig. Toby co-founded Rig after seeing how hard it was for business teams and AI to work with complex internal data. He was previously co-founder and CEO of Lingumi and CPO at Novakid.

Talk to Rig

## Check the migration before you commit to it

An engineer sends a short written view of the scope, systems and risks, then offers a call to go through it.

[Request a migration readiness check](https://rig.so/book-demo?offer=readiness_check&from=%2Fblog%2Flooker-migration-guide)[Book a call with an architect](https://rig.so/book-demo?offer=architect_call&from=%2Fblog%2Flooker-migration-guide)

---

- [This page as HTML](https://rig.so/blog/looker-migration-guide)
- [Site map for language models](https://rig.so/llms.txt)
- [API and agent documentation](https://rig.so/developers)
