Business Intelligence & Dashboards
Metabase
Rest of world Report an error0–5 in half steps. 5 means the rubric's top anchor is met on the evidence.
by Metabase, Inc. · www.metabase.com
Compare with Cluvio → Report an error on this page Is this your product? →
Link to this page
Add ?theme=dark to the image address for dark backgrounds.
The badge shows the category and the month of the latest evaluation, and EU-Made where
that is the product’s provenance. It never shows a score.
Metabase at a glance
- Vendor
- Metabase, Inc.
- Headquarters
- Not stated on the vendor’s pages as of 1 Nov 2026
- Data hosting
- Not stated on the vendor’s pages as of 1 Nov 2026
- Free plan
- Not stated on the vendor’s pages as of 1 Nov 2026
- Starting price
- “Starter $100 / month” (vendor’s words) source, captured 1 Nov 2026
- Panel verdict
- The AI panel rates Metabase 3.5/5, the mean of 6 judges. Metabase is a business intelligence product that queries 18 official database and warehouse connectors live, with a no-SQL query builder and a SQL editor. how scores are made
- As of
- Vendor pages captured 1 Nov 2026; evaluated by the panel 1 Nov 2026. all sources
Read this page as one judge. Each weighs the same scores by what they care about.
The panel's verdict
Metabase is a business intelligence product that queries 18 official database and warehouse connectors live, with a no-SQL query builder and a SQL editor. It scored highest on sharing and embedding (7 to 9), where server-signed embeds, locked parameters for per-viewer data isolation and a React SDK are documented, and on governance and portability (7 to 8), where permissions reach row and column level alongside an audit log, a REST API and an AGPL self-hosted edition. Self-service and modelling held at 7 to 7 on a semantic layer whose metric edits propagate to every question. The splits are data connectivity (4 to 7) and sovereignty (5 to 8). Connectivity divided on the same fact: judges with database stacks saw 18 official connectors; judges needing ERP, CRM or accounting connections found only tool-by-tool guides for Stripe, HubSpot and Shopify. Sovereignty divided between self-hosting with an air-gapped option and a Europe (Frankfurt) cloud region on one side, and the Metabase, Inc. contracting entity under California law, with no public information on subprocessors or a data processing agreement, on the other.
Speaks for it
- Sharing and embedding drew the bench's highest scores (7 to 9) on server-signed embeds, locked parameters for per-viewer isolation and a React SDK, with embedding "From $575 / month" on Pro and Enterprise.
- Self-service and modelling scored 7 to 7, with metric edits propagating immediately to every question that uses them.
- Governance and portability scored 7 to 8, with permissions to row and column level, an audit log and a documented REST API with a downloadable OpenAPI 3.1.0 specification.
- Pricing transparency scored 7 to 8, with Starter at $100 / month, Pro at $575 / month and every user who signs in counted for billing.
Held against it
- Data connectivity drew the bench's lowest score (4 to 7), as the catalog covers databases and warehouses while Stripe, HubSpot and Shopify appear as tool-by-tool guides.
- Sovereignty scored 5 to 8 against a Californian contracting entity, with no public information on subprocessors or a data processing agreement.
- We found no public information on pivot tables, mobile or TV display, print-ready output or chart accessibility.
- Metrics work only in the query builder and are not available in joins or nested questions.
- Enterprise is a $20,000 / year floor and the air-gapped edition carries an unspecified additional fee.
Best for
- You embed customer-facing analytics in your own product and need each customer to see only their own rows.
- Your data sits in mainstream databases and warehouses and you want live queries rather than ingestion.
- You need one official definition per metric, applied everywhere it is used and updated in place when it changes.
- You want to budget Starter or Pro from the published pricing page before talking to sales.
Avoid if
- Your data sources are business applications such as CRM, accounting or ERP systems rather than databases and warehouses.
- You need metrics to work inside joins or nested questions rather than single-source query-builder questions.
- Your access rules must also govern native SQL access, which the documented limits state bypasses row and column security.
- You require a European contracting entity; the terms of use name Metabase, Inc. under California law.
The scores
| Criterion | Panel median |
|---|---|
| Data sources, connectors & refresh | 3.5 of 5 |
| Dashboards, charts & interactivity | 3.5 of 5 |
| Self-service analysis & shared metrics | 3.5 of 5 |
| Sharing, scheduled reports & embedding | 4 of 5 |
| Access control, administration & portability | 4 of 5 |
| European sovereignty | 3.5 of 5 |
| Pricing transparency | 4 of 5 |
Data sources, connectors & refresh
panel disagrees
Show reasoningHide reasoning
How this is scored
Which data the product can reach and how it stays current: native connectors to databases and warehouses (PostgreSQL, MySQL, SQL Server, BigQuery, Snowflake), to business applications (accounting, ERP, CRM, shop, help desk, ad platforms) and to files and spreadsheets (Excel, CSV, Google Sheets), live queries versus imported extracts, scheduled and incremental refresh with its minimum interval, joining data from several sources, light data preparation, access to databases behind a firewall (gateway, SSH tunnel, IP allow-list), and a generic REST or custom connector for what is not listed.
0 — Not a reporting product, or data can only be typed or pasted in by hand.
3 — Spreadsheet and file uploads plus a handful of named connectors; refresh schedules, live queries and blending sources are basic or not described.
5 — A published connector list covering the common databases, warehouses, spreadsheets and several named business applications, scheduled refresh with its interval stated, live queries or extracts at the customer's choice, data from two sources combined in one report, and a documented way to reach a database behind a firewall.
8 — Hundreds of maintained connectors or a full SQL layer over every major warehouse, incremental refresh and caching explained, a built-in data preparation or modelling step, a gateway or agent for on-premise sources, a generic REST or custom connector SDK, and connector limits (rows, API calls, history) stated per plan.
10 — Connectivity an organisation can standardise on: every source reachable live or by extract with the trade-offs documented, refresh down to minutes with lineage from report back to source, source changes detected and reported, and connector behaviour, quotas and failure handling published.
The Finance Controller
The published connector list names eighteen maintained database and warehouse connectors plus CSV upload and Google Sheets sync on cloud, all queried live with the data staying in our database, an SSH tunnel for anything behind our firewall, and joins, models and scheduled model refresh for preparation. What holds me back: the catalog covers databases and warehouses rather than accounting, ERP or CRM systems — Stripe, Google Analytics, HubSpot and Shopify appear as guide topics — and I found no public information on a generic REST connector, stated refresh intervals, or per-plan connector limits. 6 3 15 5
The Managing Director
Strong live connectivity to 18 official database and warehouse connectors plus CSV upload and Google Sheets sync on cloud plans, with SSH tunnels and SSL modes documented for secured databases. But for a company like mine running HubSpot, accounting and a help desk, the pages show only tool-by-tool guides rather than ready-made connectors, and we found no public information on a generic REST connector or a stated minimum refresh interval. 6 5 15 3
The Data Analyst
Eighteen maintained connectors cover the warehouses and databases in my stack — PostgreSQL, Snowflake, BigQuery, Redshift, SQL Server among them — with SSH tunnels, five SSL modes and client certificates for databases behind a firewall, and queries run live so nothing is ingested; CSV upload and Google Sheets sync on cloud plans cover files. The 250-plus business-application pages are setup guides rather than native connectors, and I found no published connector limits per plan. 6 5 15 1
The IT Administrator
Eighteen maintained connectors cover PostgreSQL, SQL Server, MySQL, Snowflake, BigQuery, Redshift, Databricks and more, with eleven community drivers that work only when self-hosting; queries run live against our databases and nothing is stored unless we enable caching, which is the architecture I want for the ERP system. An SSH tunnel, five SSL modes up to verify-full with client certificates, and IP allowlisting are documented, joins and multi-level aggregation combine sources in one question, and models can be persisted and refreshed on a schedule we define. We found no public information on a generic REST or custom connector SDK or per-plan connector limits, and the integrations catalog offers tool-by-tool guides for Stripe, Google Analytics, HubSpot and Shopify rather than native connectors to business applications. 6 5 15 3 7 8
The SaaS Product Lead
Eighteen official connectors cover every database my fleet platform would need — PostgreSQL, SQL Server, Snowflake, BigQuery, Databricks — queried live so reports are never stale, plus CSV upload, Google Sheets sync on cloud plans, SSH tunnelling and a transforms layer for light preparation. Eleven community drivers extend the list but only when self-hosting, and we found no public information on a generic REST connector, incremental refresh, or per-plan connector limits. 3 5 6 7 15
The Skeptic
Eighteen named official database and warehouse connectors plus eleven community ones (self-hosting only, unsupported) are published, queries run live against the source — "Metabase doesn't ingest it, store it, or do refreshes" — and SSH tunnels, five SSL modes and a transforms-based modelling step are documented. But I found no connector to any business application, only an integrations catalog of "tool-by-tool guides" for Stripe, HubSpot and Shopify which are guides, not connectors; Google Sheets syncs only on Cloud plans with storage; and we found no public information on a generic REST or custom connector SDK, or on per-plan connector limits. 1 3 5 6 15 17
Dashboards, charts & interactivity
Show reasoningHide reasoning
How this is scored
What can be built and how readers use it: the range of chart types, tables and pivot tables, maps, KPI tiles with targets and comparisons, dashboard layout, filters and date controls that apply across a dashboard, drill-down and drill-through, cross-filtering by clicking a chart, templates for common sources, mobile and TV display, theming to the company's design, number formatting and localisation (decimal comma, currencies, German date formats), and accessibility of the result.
0 — No charts or dashboards; the product only stores or moves data.
3 — Standard bar, line and pie charts and a table on a dashboard; filters, drill-down and mobile display are basic or not described.
5 — A broad chart library with pivot tables, maps and KPI tiles with comparison to a previous period or target, dashboard-wide filters and date ranges, drill-down, templates for common sources, a mobile view, and number and date formats localised.
8 — Cross-filtering and drill-through between dashboards, custom visuals or a chart SDK, conditional formatting and annotations, dashboards themed to a brand, TV and presentation modes, pixel-exact or paginated reports for print, and accessibility of charts documented.
10 — Dashboards that a whole organisation reads daily: every interaction (filter, drill, cross-filter, parameter) available to readers without editor rights, performance on large data stated, accessibility conformance (WCAG) published, and every chart reproducible from a documented query.
The Finance Controller
Drill-through, cross-filtering, drag-to-filter and custom click behaviour are documented in real depth, with dashboard-wide filters, date ranges down to quarter and relative dates, goal lines and progress bars, and theming to our own colours on Pro and Enterprise. But I found no public information on pivot tables, chart templates, a mobile view, print-ready paginated output for the advisory board, or accessibility — and as the man who prints the management report, that silence matters. 17 16 20 2
The Managing Director
Impressive interactivity: dashboard, header and card-level filter widgets with date ranges and relative dates, drill-through menus, cross-filtering, drag-to-filter on time series, goal lines and progress bars, plus brand theming with colours and fonts on Pro and Enterprise. We found no public information on a mobile view, templates for common sources, pivot tables or chart accessibility. 16 17 20 21
The Data Analyst
Dashboard filters work at dashboard, header and card level with date-range and relative-date operators and multi-select, drill-through opens from any chart point, column header or table cell with break-outs and zoom, cross-filtering and drag-to-filter are documented, and goal lines and progress bars put targets on KPI charts; brand theming of colours and fonts is on Pro and Enterprise. I found no public information on pivot tables, mobile or TV display, or chart accessibility. 16 17 2 20 21
The IT Administrator
Interactivity is documented in depth: clicking a chart, column header or table cell opens a drill-through menu with Filter by this value, See these records, Break out by and Zoom in, alongside cross-filtering, drag-to-filter on time series, custom click behaviour to other questions or dashboards, and dashboard-wide filter widgets of date, location, number, text and boolean types with defaults, required values and per-user scope. Charts evidenced include bar, line and area series with goal lines, histograms, progress bars against a goal, and maps with zoom-in. We found no public information on the breadth of the chart library, pivot tables, mobile or TV display, templates, or accessibility of the charts. 17 16 21 20 2
The SaaS Product Lead
Dashboard filters are documented in real depth — date ranges and relative dates, location, multi-select, required filters with defaults at dashboard, header and card level — and interactivity goes well beyond basics: cross-filtering, drill-through menus with break-out and zoom, custom click behaviour to other dashboards, drag-to-filter, plus brand theming with custom colours and fonts on Pro. We found no public information on pivot tables, mobile or TV display, templates, or accessibility of charts. 2 16 17 20
The Skeptic
Dashboard-wide filters with six widget types, required and default values, cross-filtering and an unusually detailed drill-through menu are documented, along with maps, histograms, goal lines, progress bars, drag-to-filter ranges and light/dark theming with colour and font customisation on Pro and Enterprise. But we found no public information on pivot tables, KPI tiles with period comparison, mobile or TV display, conditional formatting, chart annotations or accessibility, and localisation appears only as an embed locale setting. 2 16 17 20 21
Self-service analysis & shared metrics
Show reasoningHide reasoning
How this is scored
Whether people outside the data team can answer their own questions and whether everyone gets the same answer: a visual query builder, a SQL editor for analysts, calculated fields and formulas, a semantic layer or data model where metrics such as revenue or churn are defined once and reused, metric definitions under version control (Git, dbt integration), questions in natural language with the generated query shown, forecasting and anomaly detection with the method stated, and notebooks or Python for deeper analysis.
0 — Dashboards can only be viewed; no one can ask a new question of the data.
3 — Charts are configured from a field list and simple calculated fields; shared metric definitions, SQL and a query builder are basic or not described.
5 — A visual query builder for business users and a SQL editor for analysts, calculated fields and formulas, saved questions reused across dashboards, a data model or dataset layer where joins and metrics are defined once, and natural-language questions that show the query they ran.
8 — A semantic layer with governed metrics and dimensions reused across every report, definitions in code under Git version control or synced with dbt, certified or endorsed datasets, forecasting and anomaly detection with the method stated, and Python or notebook analysis beside the dashboards.
10 — Self-service that stays consistent at scale: every metric defined once and traced to every chart that uses it, changes reviewed and tested before they reach readers, AI answers grounded in the semantic layer with their limits documented, and the model exportable in an open format.
The Finance Controller
This is the one definition of contribution margin I am after: a semantic layer where metrics and measures are defined once, verified on Pro and Enterprise, synced to Git, surfaced in the query builder, and updated in every question the moment the definition changes — beside a no-SQL query builder, a SQL editor and custom expressions. It falls short of the top because metrics do not work across joins or nested queries and live only in the query builder, and I found no public information on forecasting, anomaly detection with a stated method, or Python and notebook analysis. 18 19 2 3
The Managing Director
A point-and-click query builder that needs no SQL, custom expressions like CountIf, a SQL editor for specialists, and a semantic layer in Data Studio where metrics define the official way to calculate important numbers — reused across questions, verifiable on Pro and Enterprise, with measures on published tables synced to Git and a plain-English AI assistant on all plans. We found no public information on forecasting, anomaly detection or notebook analysis. 1 3 18 19 5
The Data Analyst
The point-and-click query builder needs no SQL while the SQL editor stays for me, custom expressions act like spreadsheet formulas, and metrics give one official definition — edit it and every question picks up the new calculation — with verification on Pro and Enterprise and measures on published tables synced to Git. Metabot takes plain-English questions on all plans, though I found no public information showing the generated query; metrics are documented as query-builder only and bound to a single data source, and forecasting and Python notebooks are not described. 18 19 3 5 2
The IT Administrator
A point-and-click query builder for business users sits beside a SQL editor, with custom expressions like CountIf and SumIf, joins, multi-level aggregation, preview-as-you-build and fifteen saved versions per question. Metrics and measures form a semantic layer with official shared formulas that immediately propagate to every question using them, verification on Pro and Enterprise, a metrics explorer for breakouts, comparisons and maths between metrics, and Git sync for measures on published tables; Metabot answers questions in plain English. We found no public information on forecasting or anomaly detection methods, notebooks or Python, or whether AI answers show the query they ran. 3 18 19 5 1 2
The SaaS Product Lead
The point-and-click query builder needs no SQL, custom expressions work like spreadsheet formulas, and the semantic layer in Data Studio defines metrics and measures once with Git sync, verified metrics on Pro and a Metrics Explorer — exactly the mechanism that stops six hundred customers each inventing their own churn formula. The documented limits matter: metrics work only in the query builder and only on their original data source, and we found no public information on notebooks, Python, forecasting or anomaly detection, or whether the AI assistant shows the query it ran. 1 3 5 18 19
The Skeptic
A no-SQL query builder, a SQL editor, spreadsheet-like custom expressions and metrics whose edits propagate immediately to every question using them are documented, with measures synced to Git and metrics markable as verified on Pro and Enterprise. The fine print is what a buyer needs: metrics are query-builder only, unavailable in joins or nested questions, and we found no public information on forecasting, anomaly detection, notebooks, or whether the plain-English AI assistant shows the query it ran. 2 3 5 18 19
Sharing, scheduled reports & embedding
Show reasoningHide reasoning
How this is scored
How results reach the people who need them: sharing a dashboard with colleagues and with external parties, public or password-protected links, scheduled reports by email or chat (Slack, Microsoft Teams) as PDF, image or data, threshold alerts, comments and annotations, export to PDF, PowerPoint and Excel, embedding dashboards in an intranet, a website or the company's own software product, white-label options, and per-customer data separation for embedded use.
0 — Results cannot leave the editor; no sharing, export or embedding.
3 — Dashboards shared with other users of the account and exported to PDF or CSV; scheduled reports, alerts and embedding are basic or not described.
5 — Sharing with colleagues and by link, scheduled email reports with PDF or image, alerts when a metric crosses a threshold, export to PDF and Excel, and dashboards embedded in a website or intranet with an iframe or script.
8 — Reports delivered to Slack or Microsoft Teams, comments and annotations on dashboards, signed embedding with row-level filtering per viewer, an embedding SDK for a software product with white-label styling, client or guest portals, and an embedding licence model documented.
10 — Distribution as a product: multi-tenant embedded analytics with per-customer data isolation, theming and custom domains, end-user self-service inside the host application, every delivery channel automatable through an API, and usage of shared and embedded content reported to the customer.
The Finance Controller
Alerts go out by email, Slack or webhook on schedules from every minute to a custom Quartz cron, with goal-line and progress-bar triggers, and the embedding side is a product in itself: signed guest and SSO embeds, locked parameters so a customer sees only their own orders, a React SDK, white-labelling and multi-tenant separation, priced from $575 / month. I hold back because I found no public information on PDF or Excel export of results, on the format dashboard subscriptions deliver, or on comments and annotations. 21 20 5 7
The Managing Director
My Monday summary is achievable: alerts and dashboard subscriptions to email, Slack and webhooks, schedulable down to the minute with custom cron, plus public links. The embedding side is deep — locked parameters so a customer only sees their own rows, white-label styling, guest embeds and a React SDK — with embedding priced from $575 per month on Pro and Enterprise; we found no public information on comments or annotations on dashboards. 21 5 20 7
The Data Analyst
Alerts reach email, Slack and webhooks on schedules down to the minute with Quartz cron and goal-line triggers, with dashboard subscriptions alongside and public links for questions and dashboards. Embedding runs from guest view-only charts to full-app interactive analytics with server-signed tokens, locked parameters so a customer sees only their own orders, white-label theming and a React SDK, from $575 / month on Pro and Enterprise. I found no public information on PDF, PowerPoint or Excel export formats. 21 20 5 7 3
The IT Administrator
Alerts deliver by email, Slack or webhook on schedules down to the minute or a custom Quartz cron, with goal-line and progress-bar triggers, one-time alerts and admin bulk management, and dashboard subscriptions carry results by email and Slack. Embedding is thoroughly documented: view-only and interactive charts, guest and full-app embedding, a React SDK, locked parameters set server-side so an embed on customer 13's account page returns only customer 13's orders, per-viewer data permissions on SSO embeds, white-label styling, and embedding priced from $575 / month. We found no public information on the file formats of downloadable results, or on comments and annotations on dashboards. 21 20 5 7 23
The SaaS Product Lead
This is the closest thing to my wish list I have read: server-signed JWT embeds with locked parameters so an embed on customer 13's account page returns only customer 13's orders, guest and full-app embedding, a React SDK plus a no-code web-component path, white-label theming, custom domains, an embeddable AI chat, and alerts to email, Slack or webhooks on schedules down to the minute. The embedding licence is public — "From $575 / month" on Pro and Enterprise — and admins can see every alert and dashboard subscription in usage analytics; we found no public information on comments or annotations on dashboards, or on automating delivery channels through the API. 1 5 7 9 20 21
The Skeptic
Alerts deliver via email, Slack or webhook on schedules from by-the-minute to custom Quartz cron, dashboard subscriptions exist, and embedding is documented three ways — view-only, interactive SSO with per-viewer data isolation via server-set locked parameters, and a React SDK with white-label styling — priced "From $575 / month". But guest embeds are always view-only and carry a "Powered by Metabase" banner on OSS and Starter, row and column security does not apply to public links, and we found no public information on export to PDF or PowerPoint or on comments and annotations. 5 7 20 21 22
Access control, administration & portability
Show reasoningHide reasoning
How this is scored
What a company needs before it gives the tool to everyone, and what leaving would cost: roles and folder or workspace permissions, row-level and column-level security, single sign-on (SAML, OpenID Connect) and user provisioning (SCIM), audit logs and usage analytics, content certification and lifecycle, environments for development and production, a public REST API for administration and content, export of dashboards and definitions in a reusable format, open-source code, and a self-hosted or on-premise edition the customer runs on its own servers.
0 — A single shared login with no permissions; dashboards cannot be exported or administered.
3 — Users with view and edit rights; row-level security, single sign-on, an API and export of definitions are basic or not described.
5 — Roles and folder or workspace permissions, row-level security documented, single sign-on on at least one stated plan, an audit log, a documented REST API, and dashboards and data exportable on termination.
8 — Column-level security and data permissions inherited from the source, SAML and SCIM provisioning, usage analytics for admins, development and production environments or content promotion, dashboards and definitions as code that can be versioned and moved, and a self-hosted or on-premise edition with its own documentation.
10 — Governance and an exit that can be verified: open-source code under a named licence that runs fully self-hosted with feature parity stated, every object (data model, dashboards, permissions) exportable in a documented format another instance imports, audit logs streamed to the customer's own systems, and deletion of every copy on termination documented.
The Finance Controller
Permissions reach down to table, row and column, impersonation defers access control to the database's own roles — which our auditors will appreciate — and there is an audit log with usage analytics, a documented REST API with an OpenAPI specification, staging-to-production export with stable entity IDs, and a self-hosted AGPL edition with an air-gapped option. Short of the top: the API is explicitly not versioned, and I found no public information on SCIM provisioning, while single sign-on is documented without naming its protocol. 4 22 23 10 1 20
The Managing Director
Permissions down to database, table, schema, row and column level on Pro and above, database-role impersonation, an audit log, usage analytics, staging environments with export of configs, models and dashboards, a documented REST API with a downloadable OpenAPI 3.1.0 specification, and a self-hosted Docker and JAR edition licensed under AGPL. We found no public information on SCIM provisioning or documented deletion of every copy on termination. 4 22 23 5 8 13 1
The Data Analyst
Permissions reach row and column level on Pro and Enterprise, impersonation hands view permissions to database-defined roles, and admins get an audit log and usage analytics. A documented REST API automates administration — explicitly not versioned — staging environments export configs, models and dashboards with stable entity IDs between instances, and the AGPL-licensed open-source edition self-hosts via Docker, JAR or an air-gapped Enterprise build, so leaving means moving files, not rebuilding dashboards. I found no public information on SCIM provisioning or streaming audit logs out. 4 22 23 5 10 13 8 1
The IT Administrator
Permissions reach database, table, schema, row and column level with groups and collection controls, a comprehensive audit log with alerts on unusual activity, usage analytics, and a documented REST API with OpenAPI spec and API keys, while SSO with a provider and required 2FA arrive from Pro upward. Row and column security is documented in detail, including its own stated limits — it applies to query-builder questions, native SQL results are not secured, public links are not covered — and impersonation pushes permissions down to our database roles. The application is AGPL-licensed, self-hosting by Docker, JAR or Podman has its own documentation, and staging environments with exported configs, models and dashboards plus stable entity IDs support moving content between instances; we found no public information on SCIM provisioning or on streaming audit logs to our own systems. 4 22 23 15 5 7 10 13 8 1
The SaaS Product Lead
Permissions run from database down to table, schema, row and column, backed by row-and-column security through saved SQL views, impersonation against database roles, SSO with required 2FA and IP allowlists on Pro, an audit log, download restrictions, and staging environments with content serialization between instances. The REST API is documented with OpenAPI and API keys, and the open-source edition under the AGPL licence self-hosts via Docker or JAR with an air-gapped Enterprise option; we found no public information on SAML or SCIM provisioning. 1 4 5 8 10 13 20 22 23
The Skeptic
Permissions are documented down to database, table, schema, row and column, impersonation hands row-level control to database-defined roles, there is an audit log, usage analytics, content serialisation with stable entity IDs for staging-to-production moves, a documented REST API with a downloadable OpenAPI 3.1.0 specification, and an AGPL-licensed self-hosted edition with its own documentation. The pages also state the limits plainly, which I respect: native SQL access bypasses row and column security, the API is explicitly not versioned, and account termination "may result in destruction of any Content" — the exit is the self-hosted edition and serialisation, not the cloud contract. 4 5 8 10 13 20 22 23
European sovereignty
panel opinion
panel disagrees
Show reasoningHide reasoning
How this is scored
Who the contracting entity and the parent company are, where the connected business data is queried, cached, extracted and stored, and who the subprocessors are (hosting, query engine, AI model providers, email delivery of reports, support and payment tools). Independently sourced by the sovereignty pipeline; a BI tool reads revenue, customer and employee data from many systems, so every place a copy or cache lands counts, not only the account database. A self-hosted or open-source edition the customer runs on its own European servers counts as the customer's chain, whatever the vendor's seat.
0 — Non-EU vendor and contracting entity, hosting unstated, subprocessors unnamed, and no data processing agreement offered.
3 — A privacy policy and a data processing agreement exist, but hosting location is unstated or non-EU and the subprocessors are not listed.
5 — EU hosting available and an EU contracting entity or EU representative, but parts of the chain (query engine, extracts and caches, AI features, report emails, support, payment) run through non-EU providers without a stated safeguard.
8 — EU hosting on named infrastructure, EU contracting entity, subprocessors listed with their locations, a data processing agreement published, where extracts and caches are stored stated, and security certifications (ISO 27001, SOC 2) stated — or a self-hosted edition documented that keeps every query and cache on the customer's own servers.
10 — Sovereign end to end and evidenced: European ownership, entity, hosting and every subprocessor including AI features, or a self-hosted open-source edition with no call home, data export and deletion documented, and certifications published with their scope rather than implied.
The Finance Controller
Self-hosting is documented to the point of an air-gapped Enterprise edition, with the application under AGPL, data staying on our own servers, open-source instances that never need to call home, SOC 2 Type II and SOC 1 stated, and a Europe (Frankfurt) cloud region available — that is the customer's own chain, and it is where our data would stay. The counterweights are a Californian contracting entity under California law, and I found no public information on a data processing agreement, a subprocessor list with locations, or where cloud-side caches land beyond the statement that nothing is stored unless results caching is enabled. 11 8 13 7 9
The Managing Director
Cloud hosting is available in Europe (Frankfurt) with a documented region-switch procedure, a stated GDPR posture with SOC 2 Type II, and a self-hosted edition where the data stays on our own servers. But the contracting entity is Metabase, Inc. under California law, and we found no public information on a data processing agreement, an EU representative or a subprocessor list with locations — the security page describes vendor oversight without naming the vendors. 13 9 11 8
The Data Analyst
The self-hosted edition is documented to keep every query on our own servers — none of the data returned by the database is stored on the vendor's side unless results caching is enabled — with an air-gapped Enterprise deployment available, and the cloud offers a Europe (Frankfurt) region. The contracting entity is Metabase, Inc. in California under California law; SOC 2 Type II and SOC 1 are stated, and I found no published subprocessor list or data processing agreement. 11 8 9 13 7
The IT Administrator
The contracting entity is Metabase, Inc. under California law, but Metabase Cloud offers a Europe (Frankfurt) region, each cloud customer runs in its own containers with no shared clusters, and unless results caching is enabled none of the data returned by our database is stored on the vendor's side. The open-source edition we would run keeps everything on our own servers — "Your data stays on your servers—and in your control. We never look at or copy your data. Period." — with an air-gapped edition available on Enterprise for an additional fee. Subprocessors are reviewed before use and at least annually but are not listed with their locations on these pages, we found no public information on a published data processing agreement, and we found no public information on where the paid AI service processes data. 13 9 11 8 7
The SaaS Product Lead
The contracting entity is Metabase, Inc. under California law, but cloud hosting offers a Frankfurt region with per-customer containers and the pages state that unless results caching is enabled none of the data returned by the database is stored on the vendor's side, while the documented self-hosted edition keeps everything on the customer's own servers — which, run on European infrastructure, becomes our own chain. Subprocessors are reviewed internally but not publicly listed, and we found no public information on a data processing agreement. 7 9 11 12 13
The Skeptic
The contracting entity is Metabase, Inc. under California law, cloud hosting offers a Europe (Frankfurt) region, and the security page states query results are not stored "on our end" unless caching is enabled, while self-hosting keeps data on the customer's own servers and open-source instances "never need to call into us at all". Still, I found no public information naming the subprocessors or on a data processing agreement, and the AI service priced at $3.75 per 1m tokens sends its prompts somewhere the captured pages never name. 7 9 11 13
Pricing transparency
Show reasoningHide reasoning
How this is scored
Whether a buyer can compute the real annual cost from public pages alone: price per creator or editor seat and whether viewers are free or paid, capacity or usage pricing (rows, refreshes, queries, AI credits), which features (row-level security, SSO, embedding, API, more connectors, faster refresh) sit behind a higher tier or "contact sales", the price of embedding for external users, the self-hosted edition's licence, the free tier and its limits, minimum term and setup.
0 — No public prices at all; every customer is a sales call.
3 — A starting price or a free tier exists, but how cost scales with viewers, data volume or features is unclear.
5 — Tiers are public with their seat price and main inclusions, but at least one commonly needed piece (viewer seats, row-level security, SSO, embedding, extra connectors or refresh frequency) is unpriced or "on request".
8 — Every tier priced publicly with its scaling rule (creators, viewers, capacity, rows), viewer pricing stated, add-ons and usage overage priced, embedded and self-hosted pricing stated, minimum term and VAT treatment stated.
10 — Complete price computability: annual cost derivable for a given number of creators and viewers, data volume and feature set, with every add-on, embedding, setup, support and overage fee and the cancellation and data-return terms published.
The Finance Controller
From the pricing page alone I can budget the year: Starter at $100 / month with the first five users then $6 per user / month, Pro at $575 / month with ten then $12 per user / month, storage at $2 for 1M rows with 1M free, transforms at $0.01 per run after 1,000 included, AI at $3.75 per 1m tokens after 1m included, embedding from $575 / month, a 10% yearly discount, prices excluding taxes, and per-user billing prorated. What stops full computability: Enterprise is custom above its $20,000 / year floor, the air-gapped edition carries an additional fee without a figure, and I found no public information on minimum term or setup costs. 7
The Managing Director
I can compute my annual cost from the pricing page: Starter at $100 per month with the first 5 users included and $6 per user per month after, Pro at $575 per month with 10 included and $12 per user per month, Enterprise starting at $20,000 per year, a free open-source edition, a 10% yearly saving, prorated billing, and priced overages for transforms ($0.01 per run after 1,000 included), AI tokens ($3.75 per 1m tokens) and storage ($2 for 1M rows, first million free). We found no public information on the minimum term, self-hosted paid licence prices or the air-gap fee, and every person who signs in counts as a billed user. 7 2 3
The Data Analyst
Every tier is public: Starter $100 / month with the first 5 users included and $6 per user / month after, Pro $575 / month with the first 10 users and $12 per user / month, Enterprise starting at $20,000 / year, and the open-source edition free; usage is priced too — $2 for 1M rows of storage with 1M free, transforms $0.01 per run and advanced transforms $0.02 per run after 1,000 included, AI $3.75 per 1m tokens after an included million. Prices exclude taxes, yearly billing saves 10%, everyone who signs in is a billed user and billing is prorated; the air-gapped deployment carries an unspecified additional fee and I found no published minimum term. 7 3
The IT Administrator
Every plan is public: Starter at $100 / month with $6 per user / month after the first five included, Pro at $575 / month with $12 per user / month after the first ten included, Enterprise starting at $20,000 / year, storage at $2 for 1M rows beyond 1M free, transforms at $0.01 per run and advanced transforms at $0.02 per run after 1,000 included runs, AI at $3.75 per 1m tokens after 1M included, a 10% yearly saving, prorated billing, a free open-source edition, a 14-day trial and prices that exclude any applicable taxes. Everyone who signs in counts as a billed user with non-transferrable licences, which is a clear scaling rule I can budget from. Enterprise beyond its floor and the air-gapped option are priced by agreement, and we found no public information on a minimum term. 7 17 3
The SaaS Product Lead
I can put a number on the embedding licence before signing: Pro is "$575 / month" with "$12 per user / month" after the first ten included, embedding "From $575 / month", storage "starts at $2 for 1M rows" with "1M stored rows are always free", transforms at "$0.01" and advanced at "$0.02" per run, AI at "$3.75 per 1m tokens" with a million included, "Yearly Save 10%", and everyone who signs in — including via SSO — is billed as a user, so the cost of six hundred customer viewers is computable. "Prices exclude any applicable taxes"; Enterprise "Price starts at $20,000 / year" without itemisation, and we found no public information on minimum term or the air-gap fee. 7
The Skeptic
Starter at $100 / month with five users then $6 per user / month, Pro at $575 / month with ten users then $12 per user / month, storage at $2 for 1M rows after the first million free, transforms at $0.01 and advanced transforms at $0.02 per run, and AI at $3.75 per 1m tokens give a buyer most of the arithmetic — and the definition "Everyone who signs into your Metabase... is a user for purposes of billing" at least tells you plainly that viewers are never free. But Enterprise is only a "$20,000 / year" floor, the air-gapped edition carries an unspecified "additional fee", feature gates like row and column security and the embedding SDK sit on Pro and Enterprise, and we found no public information on minimum term or setup costs. 7 18 20 22
European sovereignty — proven facts
0 of 6 dimensions provenBuilt only from facts shown on the vendor's own pages. A dimension we could not prove is left open, not scored as zero.
| Legal entity | Not determined | — | uncited Report an error |
|---|---|---|---|
| Hosting location | Not determined ⚠ unverified | — | uncited Report an error |
| Hosting options | Not determined ⚠ unverified | — | uncited Report an error |
| Open source | Not determined ⚠ unverified | — | uncited Report an error |
| Data processing agreement | Not determined | — | uncited Report an error |
| Transfer impact assessment | Not determined | — | uncited Report an error |
Questions about Metabase
Is Metabase European?
No. whats-best.ai classifies Metabase as Rest of world.
Where does Metabase store data?
Not stated on the vendor’s pages as of 1 Nov 2026.
Does Metabase have a free plan?
Not stated on the vendor’s pages as of 1 Nov 2026.
What are the best alternatives to Metabase?
As of 1 Nov 2026, the panel rates Preset (Apache Superset) (3.5/5), Holistics (3/5) and Luzmo (3/5) highest among the other 15 Business Intelligence & Dashboards products evaluated. Ratings that look equal are rounded to half stars; the order uses the unrounded rating, then the name. The highest-rated European alternative is Luzmo (EU-Made). Ratings are the AI panel’s opinion, formed from public evidence: how scores are made. The full list, with European products first: Metabase alternatives.
Where this could be wrong
- Evidence ages. The oldest capture behind this page is from 1 Nov 2026. Vendors change pricing and policies without notice; every fact reflects its source as of the capture date shown in the registry.
- Weak sourcing — Open source, Hosting options. Not confirmed on the vendor’s own pages as captured.
- Weak sourcing — Hosting location. The page lists cloud regions without naming a default, and the same page says Metabase keeps as much data as possible on the customer's own servers; self-hosting anywhere is also offered.
- AI can misread a source. Extraction and judgement are automated; a citation guarantees traceability, not infallibility. If something here is wrong, say so — no account needed, every report is decided within 5 business days, and accepted corrections are published.
What we left out
A claim that does not survive our checks costs us the claim, not the page. This is what was taken off this one.
- 12 product facts could not be confirmed on the vendor’s page as captured and were left out of this page and of the panel’s material. Know more? Tell us
- 9 integrations facts could not be confirmed on the vendor’s page as captured and were left out of this page and of the panel’s material. Know more? Tell us
- 4 legal facts could not be confirmed on the vendor’s page as captured and were left out of this page and of the panel’s material. Know more? Tell us
- 4 support facts could not be confirmed on the vendor’s page as captured and were left out of this page and of the panel’s material. Know more? Tell us
- 2 pricing facts could not be confirmed on the vendor’s page as captured and were left out of this page and of the panel’s material. Know more? Tell us
- 2 sovereignty dimensions could not be confirmed on the vendor’s own pages and are shown as unknown. Know more? Tell us
- 6 of the readings below were written against an earlier fact sheet — a fact has been corrected, added or pulled since. Until the panel next runs on this product you are reading the older judgement. Know more? Tell us
Sources (24)
The pages every claim on this page was read from — each one checked, dated, and kept verifiable.
- 1 Vendor homepage www.metabase.com Checked 1 Nov 2026 Details →
- 2 Features page www.metabase.com Checked 1 Nov 2026 Details →
- 3 Features page (query builder) www.metabase.com Checked 1 Nov 2026 Details →
- 4 Features page (permissions) www.metabase.com Checked 1 Nov 2026 Details →
- 5 Features page (embedded analytics) www.metabase.com Checked 1 Nov 2026 Details →
- 6 Integrations page (data sources) www.metabase.com Checked 1 Nov 2026 Details →
- 7 Pricing page www.metabase.com Checked 1 Nov 2026 Details →
- 8 Documentation (self-hosting) www.metabase.com Checked 1 Nov 2026 Details →
- 9 Documentation (cloud region) www.metabase.com Checked 1 Nov 2026 Details →
- 10 API documentation www.metabase.com Checked 1 Nov 2026 Details →
- 11 Security / trust page www.metabase.com Checked 1 Nov 2026 Details →
- 12 Privacy policy www.metabase.com Checked 1 Nov 2026 Details →
- 13 Terms of service www.metabase.com Checked 1 Nov 2026 Details →
- 14 Data sources, connectors & refresh — found from sitemap www.metabase.com Checked 1 Nov 2026 Details →
- 15 Data sources, connectors & refresh — found from sitemap www.metabase.com Checked 1 Nov 2026 Details →
- 16 Dashboards, charts & interactivity — found from sitemap www.metabase.com Checked 1 Nov 2026 Details →
- 17 Dashboards, charts & interactivity — found from sitemap www.metabase.com Checked 1 Nov 2026 Details →
- 18 Self-service analysis & shared metrics — found from sitemap www.metabase.com Checked 1 Nov 2026 Details →
- 19 Self-service analysis & shared metrics — found from sitemap www.metabase.com Checked 1 Nov 2026 Details →
- 20 Sharing, scheduled reports & embedding — found from sitemap www.metabase.com Checked 1 Nov 2026 Details →
- 21 Sharing, scheduled reports & embedding — found from sitemap www.metabase.com Checked 1 Nov 2026 Details →
- 22 Access control, administration & portability — found from sitemap www.metabase.com Checked 1 Nov 2026 Details →
- 23 Access control, administration & portability — found from sitemap www.metabase.com Checked 1 Nov 2026 Details →
- 24 Code repository (open-source licence) — found from the homepage github.com Checked 1 Nov 2026 Details →