Migrating from Looker to Superset for Australian Government Organisations

Table of Contents

1. Why Australian Government Agencies Are Moving Off Looker
2. Scoping Your Migration: What’s in Your Looker Instance?
3. Governance and Compliance in a Government Context
4. Cost Benchmarks: From Per-Seat Licenses to Open-Source Freedom
5. Technical Migration Playbook: Step by Step
   • 5.1 Stand Up Your Superset Environment
   • 5.2 Export Your Looker Assets
   • 5.3 Translate LookML to Superset’s Semantic Layer
   • 5.4 Recreate Dashboards and Validate Data
6. The Cutover Pattern: Running in Parallel to De-Risk Go-Live
7. Post-Migration Optimisation and Long-Term Governance
8. Summary and Next Steps

Why Australian Government Agencies Are Moving Off Looker

Australian government organisations are under mounting pressure to modernise their data stacks. The driving force is not just cost—though that’s a major factor—but a fundamental shift in how the public sector consumes technology. The federal government’s whole-of-government cloud mandate, effective July 2026, explicitly requires agencies to prioritise cloud technologies for IT modernisation. This mandate has sparked extensive migration warnings across the sector, and business intelligence is squarely in the crosshairs.

Looker, long a mainstay of agency analytics, comes with a licensing model that scales linearly with users. For departments with hundreds or thousands of report viewers, the per-seat cost quickly spirals into seven figures annually. When you add the overhead of managing LookML models and the dependency on Google Cloud infrastructure, many agencies begin to question whether the value justifies the price. That’s where Apache Superset enters the conversation.

Superset is an open-source data exploration and visualisation platform that offers a no-cost alternative to Looker without sacrificing enterprise-grade features. It supports a wide range of databases, includes a powerful semantic layer, and can be self-hosted on sovereign Australian infrastructure—a critical requirement for government workloads handling PROTECTED or OFFICIAL: Sensitive data. For Australian mid-market and enterprise operators, the shift from Looker to Superset isn’t just a cost play; it’s a strategic move toward greater control, data sovereignty, and alignment with whole-of-government digital strategies. PADISO’s platform engineering services across Australia have been purpose-built to help agencies execute this transition with confidence.

If your agency is grappling with skyrocketing BI costs or a mandate to exit proprietary platforms, this guide provides a complete migration playbook tailored to the unique governance, compliance, and operational realities of Australian government organisations. You’ll learn how to scope, execute, and validate a Looker-to-Superset migration while maintaining audit readiness and IRAP alignment. Whether you’re a departmental CTO or a program director, the following sections will give you an actionable blueprint.

Scoping Your Migration: What’s in Your Looker Instance?

Before you provision a single Superset container, you need a detailed inventory of everything in your current Looker environment. Government BI landscapes tend to accumulate years of ad hoc explores, dashboards, and scheduled deliveries. Without a thorough scoping exercise, you risk losing critical reports or disrupting downstream business processes. A Fractional CTO & CTO Advisory service in Canberra can help you build a structured catalogue and manage stakeholder expectations throughout the scoping phase.

Cataloguing Looker Assets

Start by crawling your Looker instance. Extract a list of all:

• Looks (saved queries)
• Dashboards and dashboard filters
• Explores and models
• View files and their LookML definitions
• User roles and access groups
• Schedules (email alerts, webhooks, external integrations)
• Connected data sources (BigQuery, PostgreSQL, AWS Redshift, etc.)

Looker’s API and the LookML runtime can export the metadata you need. For extra certainty, pull a full system activity dump to identify which content is actually used versus which has been abandoned. This usage data will let you triage the migration into waves—prioritising high-usage, business-critical assets first while deprecating unused ones.

Assessing Complexity and Dependencies

Not all Looker artifacts are created equal. Dashboards that rely on heavy LookML-derived tables, extensive liquid parameters, or custom JavaScript visualisations will require more translation effort. Government departments often embed Looker dashboards into intranet portals or citizen-facing services; these integrations need a careful migration plan that updates iframe URLs, authentication tokens, and session management.

You’ll also want to map data source connectivity. If you’re moving from Looker-hosted instances on Google Cloud to an Australian-sovereign Superset deployment, you may need to reconfigure network paths, VPC peering, or hybrid links. The Platform Development in Canberra team specialises in sovereign cloud architectures that meet IRAP and PROTECTED requirements while connecting seamlessly to existing government datasets.

Setting Realistic Timelines

A typical government Looker-to-Superset migration for 50–200 users spans 8–16 weeks, depending on complexity. Break the project into two phases: a pilot migration (2–4 weeks) that covers a high-volume yet self-contained set of dashboards, and a full rollout (6–12 weeks) that iterates on lessons learned. Contingency buffers are essential—data pipelines, security reviews, and UAT cycles inside government often consume more time than in the private sector.

Governance and Compliance in a Government Context

For Australian government agencies, governance isn’t a nice-to-have; it’s a precondition for go-live. Superset, being open source, does not come with a compliance badge out of the box—you have to architect it in. The good news is that Superset’s flexibility makes it straightforward to layer on the controls required by the Protective Security Policy Framework (PSPF), the Information Security Manual (ISM), and agency-specific data-sharing agreements.

Mapping Looker Roles to Superset’s RBAC

Looker’s permission model is built around model sets, permission sets, and roles. Superset has a granular role-based access control (RBAC) system that can replicate most Looker permission structures. However, Superset’s concept of a “dataset” and its row-level security (RLS) implementation differ from Looker’s. You’ll need to:

• Reconstruct user groups and data access policies in Superset using its FAB (Flask AppBuilder) security framework.
• Apply RLS filters through Superset’s SQL Lab or via a custom security manager if dynamic filtering is needed.
• Test cross-domain restrictions rigorously—government datasets often have complex need-to-know boundaries (e.g., separate data silos for different program areas).

For agencies using Vanta for audit readiness, PADISO’s team can help configure Superset controls that align with SOC 2 and ISO 27001 evidence collection, even though the platform itself is not certified.

Hardening Superset for Government Workloads

A production government Superset deployment must meet the same security standards as any other critical system. That means:

• TLS 1.2 or higher for all data in transit, with certificates managed through agency PKI.
• OAuth 2.0 or SAML single sign-on (SSO) integrated with Entra ID (Azure AD) or a government-issued IdP. Many agencies have already standardised on Azure AD, which can be wired into Superset using the FAB SSO integration. This comprehensive guide on securing Superset walks through the authentication and encryption setup.
• Network segmentation so that Superset’s application layer sits in a secured subnet, and data warehouses reside in protected VPCs with strict firewall rules.
• Audit logging of all queries, dashboard views, and data exports. Superset’s event logging can forward logs to a central SIEM; supplement with database-level audit trails for data modification events.

For agencies governed by the ISM, the deployment must align with PROTECTED controls. This typically means hosting Superset in an Australian-assured cloud environment (e.g., within an accredited AWS Region or on government-managed infrastructure). PADISO’s Darwin platform development team has deep experience with edge and intermittent-connectivity pipelines, which is useful for agencies with remote operational sites.

Data Residency and Sovereignty

Australian government data, especially when classified as OFFICIAL: Sensitive or above, must reside in Australian data centres. Superset’s backend and metadata store (PostgreSQL or MySQL) should be hosted in-country, and all database connections must route through local gateways. If you’re using a cloud-native data warehouse like Google BigQuery or AWS Redshift, ensure the instances are launched in Sydney, Canberra, or Melbourne regions. The Platform Development in Melbourne team can help modernise regulated monoliths while maintaining data localization.

Cost Benchmarks: From Per-Seat Licenses to Open-Source Freedom

The most compelling reason for government agencies to migrate is often financial. Looker’s quoted list price can exceed AUD 3,000 per creator seat per year and AUD 600 per viewer seat. For an agency with 50 creators and 500 viewers, that’s AUD 450,000 annually in licensing alone—before you account for underlying infrastructure. Superset, by contrast, has zero licence fees. The entire cost shifts to the infrastructure you need to run it.

Estimating Total Cost of Ownership (TCO)

A typical Superset deployment for a mid-sized government agency runs on:

• Three virtual machines (or Kubernetes pods) for web servers, with a load balancer in front.
• A managed PostgreSQL database for metadata (preferably multi-AZ for high availability).
• Optional Redis for caching and Celery for asynchronous tasks (dashboard email reports, long-running queries).

In Australian cloud regions, this stack costs roughly AUD 4,000 to AUD 12,000 per month, depending on redundancy and data volume. That’s total, not per seat. Even at the high end, the annual infrastructure cost is around AUD 144,000—a fraction of the Looker licence bill. Adding specialist platform engineering or a Fractional CTO & CTO Advisory in Sydney to oversee the migration can further accelerate ROI, typically paying for itself within the first year solely from licence savings.

Hidden Costs to Watch

While Superset is free, migration isn’t. You’ll incur one-time costs for:

• Export/translation tooling (either custom scripts or a third-party migration accelerator).
• Superset security hardening and SSO integration.
• Developer time to refactor LookML models into Superset virtual datasets or SQL views.
• User training and change management.

Agencies that under-budget for training often see the lowest adoption rates. Budget for hands-on workshops, quick-reference guides, and a two-week parallel-run period where both Looker and Superset are live. PADISO’s platform development services on the Gold Coast have helped local health teams cut BI costs by over 60% using Superset while keeping training overhead low.

Long-Term Savings and Freedom

Beyond licence fees, Superset eliminates vendor lock-in. You’re not beholden to a vendor’s product roadmap or price increases. The platform is backed by the Apache Software Foundation and a vibrant open-source community, so improvements are continuous. When the government’s Digital Transformation Agency pushes for greater use of open-source data tools, agencies already on Superset can demonstrate compliance effortlessly.

Technical Migration Playbook: Step by Step

The technical migration spans five core phases. Each phase is designed to be reversible to a checkpoint, so if something fails, you can roll back without data loss. Throughout the process, keep a detailed runbook—government audit teams will appreciate it later.

5.1 Stand Up Your Superset Environment

First, deploy Superset in a pre-production environment that mirrors your target production setup. Many agencies choose a Docker Compose or Kubernetes deployment, following the official Superset documentation. If you’re running on AWS, the Platform Development in Perth team can assist with infrastructure-as-code that adheres to ISM controls. Key configuration steps:

• Set SECRET_KEY to a high-entropy random value.
• Configure database connection URIs with SSL certificates.
• Enable SSO via OAuth2 or SAML—test with a pilot group of users before opening up to everyone.
• Apply Superset’s recommended row-level security patches and disable unnecessary features (e.g., public dashboards).
• Set up a CI/CD pipeline for configuration changes, ideally using Pulumi or Terraform, to ensure consistency across environments.

5.2 Export Your Looker Assets

Looker does not provide a one-click export for dashboards into Superset format, but you can extract the raw definitions. Use Looker’s API to download:

• The LookML model and view files (Git-backed for most agencies).
• Dashboard YAML definitions (via GET /dashboards/{dashboard_id} with the content_type=application/json header).
• User roles and groups (via Admin API).

For any per-agency customisation (like embedded iframes), document the front-end integration points so you can update the URLs and tokens later. Google Cloud’s own Looker self-service migration documentation outlines export and import steps that, while intended for moving between Looker instances, can be adapted to extract metadata for third-party tools.

5.3 Translate LookML to Superset’s Semantic Layer

This is the most complex part of the migration. LookML is a YAML-based modelling language that generates SQL; Superset uses either virtual datasets (SQL queries saved as virtual tables) or the new Superset Semantic Layer (beta). The two approaches:

1. Virtual Datasets: Write the equivalent SQL directly in Superset. This is the most straightforward—you take the compiled SQL from Looker’s explore page and turn it into a virtual dataset. Row-level security filters must be appended in the UI or via a custom middleware.
2. Semantic Layer: If you have many complex metrics and dimensions, consider building a semantic layer using Superset’s datasets API or third-party tools like dbt. This mimics Looker’s explore_source and keeps business logic version-controlled.

A practical strategy is to start with virtual datasets for speed, then gradually refactor into a semantic layer for high-value models. For reusability, define common metrics as SQL templates in a versioned repository. A mid-market migration guide illustrates wave-based rollout techniques that work well for government teams, allowing them to trial a small set of models before a full cutover.

5.4 Recreate Dashboards and Validate Data

With semantic models in place, rebuild each dashboard in Superset. Superset’s charting library covers most Looker visualisations, but some custom JS charts won’t have exact equivalents. In those cases, build an alternative visualisation that tells the same story, or use Superset’s custom chart plugin capability if coding resources are available.

Validation is two-fold:

• SQL output: Compare aggregated query results between Looker and Superset for the same date range and filters. Automate this with a Python script that hits both APIs and compares CSV outputs.
• Visual parity: Business users must confirm that dashboards show the same KPIs and trends. This is best done during UAT, described in the cutover section.

Agencies with strict cross-verification requirements can run a data reconciliation report that flags any discrepancies. The Platform Development in Brisbane team has built automated validation suites for high-throughput data pipelines—a similar approach can compress the validation cycle.

The Cutover Pattern: Running in Parallel to De-Risk Go-Live

Government services cannot afford downtime. A parallel-run cutover pattern minimises risk while building user confidence. The approach has three stages:

Stage 1: Pilot Cohort (Week 1–2) Select a small group of power users (5–10 people) and give them access to the pre-production Superset environment. They’ll recreate their daily workflows, compare numbers, and report back. Fix any gaps before expanding the audience.

Stage 2: Production Superset in Read-Only Mode (Week 3–4) Deploy Superset to production with a de-risked configuration—perhaps behind the agency’s VPN—and invite the full user base. Instruct everyone to run their reports on both Looker and Superset. At this stage, decisions are still made from Looker, but users are building muscle memory. Set up a Slack or Teams channel for instant feedback; aim for zero critical bugs by the end of week 4.

Stage 3: Flip the Switch (Week 5) Once the production Superset instance has run without discrepancies for two consecutive business days, announce a cutover date. On that date, redirect all dashboard links to Superset, make Looker read-only, and turn off scheduled email deliveries from Looker. Retain Looker access for another two weeks as a safety net, then decommission it.

Training is the secret to a smooth cutover. Conduct at least two live walkthrough sessions and record them. A comprehensive comparison of Looker vs. Superset features can help you craft cheat sheets that map Looker features to their Superset equivalents. Pair these with a self-service knowledge base to head off Tier‑1 support tickets. PADISO’s services can include a fractional CTO to own the entire cutover communications plan.

Post-Migration Optimisation and Long-Term Governance

Migration is just the beginning. Once Looker is decommissioned, you have a golden opportunity to improve analytics performance and governance.

Performance Tuning for Government Workloads

Superset performance is largely a function of the underlying database. If you’re using Google BigQuery, optimise your slot reservations and partition/clustering strategies. For PostgreSQL or ClickHouse backends, fine-tune indexing and materialised views. Superset also benefits from:

• Caching query results with Redis, especially for dashboards that auto-refresh.
• Enabling asynchronous query execution in Superset for long-running reports.
• Using Superset’s embedded dashboard SDK to integrate analytics directly into existing government portals, reducing context-switching.

For agencies that handle large telemetry or IoT data, a high-performance backend like ClickHouse can deliver sub-second queries on billions of rows. The Platform Development in Adelaide team specialises in defence and advanced-manufacturing analytics, frequently pairing Superset with ClickHouse for mission-critical dashboards.

Embedding Superset and Citizen-Facing Analytics

Many government agencies publish public dashboards—COVID statistics, budget transparency, performance metrics. Superset’s embedded analytics capabilities let you integrate charts into any web application via iframe or API, with full control over authentication and row-level security. This approach eliminates the need for a separate public BI tool and keeps everything under one roof. The Platform Development in Sydney team has designed multi-tenant Superset architectures that serve both internal and citizen-facing dashboards from a single codebase.

Continuous Governance and Audit Readiness

With Looker gone, you must maintain the same level of governance—if not higher. Schedule quarterly reviews of:

• User access audits: remove stale accounts, recertify roles.
• RLS policy validation: ensure that sensitive data is not leaking across program boundaries.
• Superset version upgrades: the open-source project moves quickly, and security patches are frequent.

If your agency is pursuing or maintaining ISO 27001 or SOC 2, Vanta can continuously monitor Superset’s configuration and flag deviations. PADISO’s case studies include examples where fractional CTO leadership helped PE-backed companies achieve audit readiness across multiple tools, and the same discipline applies to government.

Summary and Next Steps

Migrating from Looker to Superset is a multi-faceted initiative that touches technology, governance, and culture. When done right, it eliminates per-seat licence overhead, strengthens data sovereignty, and aligns your analytics platform with the Australian government’s cloud-first mandate. The key to success is a structured approach: scope deeply, harden for compliance, translate LookML carefully, and cut over in a parallel run that builds user confidence.

Australian government organisations don’t have to navigate this alone. PADISO brings fractional CTO leadership, sovereign architecture expertise, and hands-on platform engineering to every stage of the journey. Whether you need a targeted engagement to execute the technical migration or an ongoing CTO as a Service partner to oversee your entire analytics strategy, we’ve delivered measurable AI and data ROI across the US, Canada, and Australia. For agencies based in the capital, a Fractional CTO & CTO Advisory in Canberra engagement can cut through procurement red tape and get the right architecture stood up quickly. In Melbourne, our Platform Development in Melbourne specialists routinely work with regulated insurance and health entities that face the same compliance pressures as government. From Darwin to Perth, PADISO’s Australia-wide platform engineering footprint ensures you have local expertise wherever your data resides.

To start your migration, inventory your Looker instance today and calculate your current total cost of ownership. Then reach out for a discovery call. We’ll help you build a business case that pays for itself in licence savings alone—and sets your agency up for a generation of open, secure, and performant analytics.