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. Looker is Google-Cloud-only, has no on-prem AI, and prices embedding as its own edition — Qrly runs on your infrastructure with embedded signed-JWT included, on one revenue-based licence.
No marketing fluff. Looker's developer experience is genuinely excellent and has set the bar for the whole category — here is where each tool is actually stronger when the dust settles and you look past the hero shots.
The features most teams actually evaluate when switching from Looker. Columns list the product as shipped — not as promised on a public roadmap.
From real migration conversations with engineering leaders and support directors. We are not arguing Looker is a bad product — it is a beautifully made one, and for pure internal data teams it is hard to beat on day-to-day feel. We are arguing it is missing specific things that a serious operations and customer-support stack needs, and those gaps are not easy to paper over with integrations.
Regulated industries, government agencies, defense contractors, healthcare providers and organisations under EU-only compliance all need a self-host option. Looker has deliberately chosen not to offer one — there is no on-prem build, no private-cloud edition, no data-processing agreement that moves the data inside your own perimeter. That is a product philosophy choice, not an oversight, and it is internally consistent with Looker's engineering focus.
It is also, for a growing slice of the market, a hard stop. Qrly runs on your Linux server, in your VPC, or on your Kubernetes cluster. The data never leaves your infrastructure unless you decide it should, and the same binary runs in air-gapped environments.
LookML is what makes Looker trustworthy, and it is also what makes it slow. A field that is not in the model does not exist: someone opens a branch, edits the LookML, gets it reviewed and deploys it before an analyst can group by it. That is excellent governance and genuinely poor turnaround, and it is why so many Looker shops still run a shadow layer of spreadsheets for the questions nobody wanted to file a merge request for.
Qrly puts the fast path first. An analyst builds the query visually or writes SQL, defines a calculated field from a SQL expression with server-side validation, and saves it — no branch, no deploy. Governance is available when it is wanted: an OLAP model with a dimension and measure catalogue, named hierarchies, and always-applied model filters that act as a security boundary, plus a verified-question badge with its own audit log.
A Looker alert is a threshold on a dashboard tile. It works, and it is the right tool when you already know which number matters and where the line sits. It does nothing at all about the metric nobody thought to put on a tile, which is reliably the one that moves the week before somebody notices.
Qrly ships the same threshold alerts — rows returned, goal reached, below goal, on an hourly, daily or weekly schedule, delivered to email, Slack or a webhook — and then adds the part that runs without being asked: nightly anomaly detection on recently used questions, an insights feed with trend-change, new-record, stale-data, correlation and seasonality detectors, and an Investigator agent that segments the anomaly and produces a structured root-cause document. Any insight is promoted to an alert in one dialog. All of it is in the core licence.
Looker charges a platform subscription and then a licence per person on top, split by role. That feels fair at twenty users and becomes a standing argument at two hundred, because growth itself turns into a line item — and the discussion about who really needs a seat starts competing with the discussion about what those people are doing with the data. We are not going to quote a euro figure at you; Google publishes the current one, and the shape of the model is the part that does not change.
Qrly's license is flat, tied to company revenue tier rather than headcount. The 201st user costs nothing. Neither does the 2,001st. Your analytics platform stops punishing you for hiring and starts behaving like infrastructure.
Our own list price in full, and an honest account of how Looker charges. 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.
There is nothing to extract. Your numbers live in the warehouse, not in Looker, so a migration is a rebuild of the presentation layer rather than a data move — which is why most teams are cut over inside a working week, with a parallel-run period where both systems stay online for cautious users.
Qrly connects to the same warehouse Looker queries — no third-party ETL, no paid connector, no export.
Looker's UX is genuinely best-in-class — coherent, governed, beautifully polished, and consistent precisely because every field arrived through the same model. We are not pretending otherwise, and Qrly is not trying to out-polish it. Qrly is productive in under five minutes and fast enough that users do not complain. The real question is whether that polish is enough to offset Google Cloud only, no self-host and per-seat pricing. For a growing number of teams, it is not.
Your data was never in Looker. LookML is a modelling layer over a warehouse you already own — BigQuery, Snowflake, Redshift, Postgres or SQL Server — and Qrly connects to that same warehouse directly through one of 40 connection types. Looks and dashboards are rebuilt in the visual builder, which for a typical dashboard is an afternoon; where you want a shortcut, paste the SQL Looker generated and Qrly reverse-engineers a flat SELECT into a visual definition, reporting CTEs, UNIONs and subqueries as explicit lossy-feature warnings. There is no LookML importer and we will not pretend there is one.
Yes. Qrly runs on a single Linux box, inside a private cloud, or on Kubernetes. There is no user minimum and no infrastructure team required — a single-container deployment is the default. Looker is SaaS-only, by design: there is no on-prem build, no private-cloud edition, no air-gapped distribution. For regulated industries, government, defense and EU-only compliance, that single difference is the entire conversation and usually ends it before features are discussed.
LookML is the best thing about Looker: a governed, Git-versioned model that makes every metric mean one thing across the organisation. Qrly's equivalent is the OLAP model — a star or snowflake defined over a connection with a dimension catalogue, a measure catalogue, named hierarchies, and always-applied model filters that act as a security boundary rather than a convention. It is authored in the UI instead of a Git-reviewed DSL: faster to stand up, and less rigorous to review. If a version-controlled modelling language is the reason you bought Looker, that is a fair reason to stay.
They solve different halves of the problem. LookML defines the model and compiles to SQL in one direction; an Explore is the governed surface on top of it. QQL is the query itself — a JSON document built in a visual builder and compiled to 12 SQL dialects, so the same saved question runs unchanged on Postgres and BigQuery, with around 35 filter operators covering regex, JSON path, array containment and four full-text operators. The part LookML does not offer is the round trip: paste a flat SELECT and Qrly converts it back into a visual definition, reporting anything lossy rather than dropping it silently. For time-based analysis you get PAST_N_DAYS, START_OF_MONTH and AT_TIMEZONE, a year to quarter to month to week to day to hour drill, and period-over-period comparison against the prior period and the prior year. What Qrly does not do is reconstruct the past state of a row — it queries your data, not an audit log of itself.
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. Looker charges a platform subscription plus a licence per person, split by role, with published pricing for Standard and quotes for Enterprise and Embed — take the current figures from Google rather than from us. The structural point survives whatever those figures turn out to be: a per-seat licence is a rounding error at fifty users and a budget line at five hundred, while our number only moves when your turnover does.
Self-hostable. Flat pricing. Embedded Analytics and Alert included. Made in Belgium.