Qrly vs Metabase
BI + AI in one self-hosted platform — natural-language Ask (NL→SQL) and BYO LLM (local Ollama / LM Studio or Anthropic, Gemini, OpenAI, Azure) included by default, on top of an OLAP model layer with ROLLUP, CUBE and period-over-period comparison. Metabase has no bring-your-own local LLM, and SSO, data sandboxing and white-labelled embedding sit in its paid editions.
Who wins where
No marketing fluff. Here is where each tool is genuinely stronger.
BI + AI in one self-hosted platform — flat pricing
- AI Ask (NL→SQL) with AI anomaly detection & schema descriptions
- BYO LLM — Anthropic, Gemini, OpenAI, Azure, or local Ollama / LM Studio
- Embedded analytics with signed JWT — not a paid edition
- OLAP models — ROLLUP, CUBE, time drill, prior-period compare
- Multi-tenant hierarchy — tenant, organisation, project, collection
- A vendor on the other end of the line — flat license, unlimited users
Core self-hosted business intelligence
- Both self-hostable on your own infrastructure
- Both run questions and dashboards straight against your own databases
- Both expose a REST API
- Both schedule results out to email and Slack on a cadence
- Software itself is inexpensive either way
Open source, and the easiest first hour in BI
- A genuinely free, self-hostable open-source edition
- The friendliest notebook query builder for non-SQL users
- A large driver ecosystem, official and community
- X-rays and auto-generated dashboards for a fast first look
- No licence cost at all if the free edition covers you
Feature-by-feature comparison
The features most teams actually evaluate when replacing Metabase.
| Feature | RecommendedQrlySelf-hosted · Belgium | MetabaseOpen source · paid Pro / Enterprise |
|---|---|---|
| Self-hostable on your own infra | Included | Core model |
| No per-user licence counting | Included | Free edition, unlimited users |
| Built-in customer embed portal | Included | Static embed; branding removal is paid |
| Alerts on saved questions (email, Slack, webhook) | Included | Alerts on questions, email and Slack |
| Native scheduled subscription (4 providers) | Included | Dashboard subscriptions, email and Slack |
| Visual query builder and raw SQL, round-trip convertible | SQL ⇄ visual | Notebook converts to SQL, not back |
| Azure AD + Google + LDAP + Basic simultaneously | Included | LDAP and Google free; SAML is paid |
| OIDC SSO user provisioning | Included | SAML and JWT SSO in paid editions |
| AI with on-prem option (Ollama, LM Studio) | Included | No bring-your-own local LLM |
| Multi-tenant architecture out of the box | Included | Data sandboxing scopes per tenant — paid editions |
| 40 connection types, 12 SQL dialects | Included | Large driver library |
| Flat pricing (unlimited users) | Included | Free software, sysadmin cost real |
| Productive in under 5 minutes | Included | Docker or JAR, minutes |
| EU data residency (native, not a tier) | Included | Included (self-host) |
| No paid edition required for the basics | Included | SSO, sandboxing and white-label are paid |
| Dashboards with cascade filters and live push | Included | Filters and auto-refresh, no push |
| REST API + webhooks | Included | Included |
| OLAP models with ROLLUP, CUBE, prior-period | Included | Models and summarise, no CUBE or compare |
| Verified questions with an audit log | Included | Verified content in the paid tier |
Four reasons teams leave Metabase
From real migration conversations with ops leads and infrastructure teams who have run Metabase for years.
The notebook stops where the hard questions start
Metabase's notebook editor is the best on-ramp in BI: pick a table, filter, summarise, join, done. It is also where most deployments plateau. Anything involving a grouping set, a running comparison against the same month last year, or a top-N inside each region drops out of the notebook and into hand-written SQL.
From that moment the logic lives in one person's saved question rather than anywhere the organisation can reuse it. Qrly keeps the governed path open further: a star or snowflake model over the connection carries the dimension and measure catalogue, named hierarchies and always-applied model filters as a security boundary, and queries against it get ROLLUP, CUBE and GROUPING SETS, a year to hour time drill, prior-period and prior-year comparison columns, and top-N within group. When you do drop to SQL, the editor converts a flat SELECT back into a visual definition, so the round trip costs you nothing.
Free software is not free to run
A JVM to size and patch, an application database that has to be moved off H2 before anyone will trust it, nginx, SSL rotation, staging clones, backup rotation, upgrade testing, and someone on call the morning a new query pattern puts the instance into a garbage-collection spiral. Someone on your team owns all of that, and their time is not free even if the software is.
None of that shows up on an invoice, which is exactly why it is easy to miss at the point of decision. It shows up as engineering hours instead — and we are not going to put a euro figure on yours, because you know that rate and we do not. What we can say is where the second bill lands: the paid edition you need the moment SAML SSO, data sandboxing or white-labelled embedding turns up on a security review, priced per user on top of the free core. Qrly is priced on revenue, not seats: a €15M-revenue company pays €6,300 a year — €18,900 over three years — for unlimited everything, one organisation, one installation, with all three of those included.
The features you need are in the paid edition
The free Metabase edition is real software, not a demo, and for a team of ten it is often the right answer. What it leaves out is short, predictable, and almost exactly the list a security review produces: SAML and JWT SSO, data sandboxing and row-level permissions, serialisation between development and production, white-labelled embedding without Metabase branding, and official support.
In Qrly none of that is a tier. Signed-JWT embedding, OIDC and LDAP sign-in, model filters applied on every query as a security boundary, verified questions with an audit log, a per-user daily query budget and a query governor all ship in the same deployment as everything else, on the same upgrade schedule, tested as one integrated system before they hit your server.
Support is a tier, not a default
The free edition is community-supported: a public forum, a public repository, and whatever your own team can work out on a Friday afternoon. For a small internal dashboard that is perfectly reasonable. For a dependency three departments rely on, it is a risk your head of IT already knows they are carrying — and the fix is a paid tier, not a patch.
Commercial users want a product roadmap they can plan around, security patches delivered on a predictable schedule, and a phone number to call when something goes sideways during a Friday afternoon deploy. Qrly ships with a named contact, a published support response time and a vendor that is accountable for the software it sells — the simple promise your finance team has been asking you to put in writing.
Pricing, honestly
Our own list price in full, and an honest account of what Metabase costs — the free edition and the paid ones. The Qrly column is the published schedule evaluated at €15M of revenue — the number of people using it does not move it.
Methodology, 28 August 2026. Qrly publishes a complete rate card, so the figures in our column are quoted exactly. The other column describes a pricing model rather than quoting a number: vendor list prices change without notice and several of these vendors do not publish one at all. For current figures, go to the vendor’s own pricing page.
Qrly
Annual licence · priced on your revenue · unlimited users, one organisation, one installation
- Annual licence at €15M revenue €6,300
- Per analyst / per viewer €0
- Per query, per credit, per capacity unit €0
- Embedded dashboard consumers €0
- Embedding, alerts, OIDC/SSO, row-level rules Included at every tier
- Self-hosting and bring-your-own LLM Included at every tier
Three years of licence: €18,900
That is our published schedule evaluated at €15M of revenue — 0.18% of the first €500k, then 0.12%, 0.0525% and 0.027% on each tranche above it. It does not move when you add users. The whole rate card, from the €900 floor upward, is on the pricing page.
Metabase
Free open-source edition · paid Pro and Enterprise on top
- Open-source edition Free — the licence costs nothing
- Pro and Enterprise Paid, priced per user
- What sits behind the paywall SAML/JWT SSO, data sandboxing, white-label embedding
- Metabase Cloud Hosted, per user, per plan
- Self-hosting the free edition Your servers, your upgrades, your on-call
- List prices Published by Metabase — check metabase.com/pricing
The open-source edition is genuinely free and genuinely good. The cost lands in two other places instead: the engineering time to run and upgrade it, and the paid tier you reach the first time somebody asks for SSO or row-level sandboxing. We are not going to put a euro figure on your engineers’ time — you know that rate and we do not. We will note that it becomes a per-user meter the moment you cross into Pro, and that SSO, white-label embedding and embed tokens with locked parameters are inside Qrly’s €900 floor. Qrly has no row-level security; locked parameters are the mechanism it offers.
Moving from Metabase to Qrly
Most teams are up and running on Qrly within a working week, pointing at the databases they already query.
The standard migration path
Nothing has to be extracted. Qrly connects to the same databases Metabase reads, so the migration is about moving definitions — questions, dashboards and permissions — not about moving data. There is no ETL step, no commercial migration vendor in the middle, and no export window to schedule.
- Point a connection at the same database. Add the source in Qrly, test it before saving, then run a schema sync so the visual builder and the SQL autocomplete know your tables. Credentials are AES-encrypted at rest and the pool hands out read-only connections, so the first connection is safe to make against production.
- Bring native questions across by pasting the SQL. Qrly reverse-engineers a flat SELECT into a visual definition — joins, filters, group by, having, order by with NULL ordering, limit and offset — and reports CTEs, UNION and subqueries as explicit lossy-feature warnings rather than dropping them quietly. Anything that will not convert still runs as native SQL, with template variables bound as prepared-statement parameters.
- Rebuild notebook questions in the builder. Metabase notebook steps map onto QQL almost one for one: source table, filters drawn from around 35 operators, aggregations, breakouts, having, order by, limit. Most teams rebuild faster than they would review an import, and the result is a query document that compiles unchanged to any of 12 SQL dialects.
- Model what the notebook could not express. Where questions had drifted into hand-written SQL, define a star or snowflake model instead — fact table, dimension joins, a measure catalogue, named hierarchies and always-applied model filters — and that logic becomes reusable, with ROLLUP, CUBE, time-hierarchy drill and prior-period comparison on top of it.
- Move users, then run in parallel. Point Qrly at the same directory — LDAP and OIDC are in the base product, so nobody re-registers. Keep Metabase read-only for around a month while dashboards are rebuilt and verified, then decommission the instance once the last holdout has stopped opening the old tab.
Common questions
Is Qrly a modern Metabase alternative?
How do we move our Metabase questions across?
Metabase is free — why pay for Qrly?
What happens to our Metabase drivers and API integrations?
Can Qrly keep the on-prem model I like about Metabase?
3-year cost comparison?
The full alternatives matrix
Evaluating Qrly against another tool? Jump to the detailed side-by-side.