Antvia vs
Metabase and Superset
Both are genuinely good products with no licence fee, and for the right team they are the correct answer. The question is which costs you are moving rather than removing.
Who arrives here
Somebody on the team said we could just self-host, it is free, and they are right about the licence. The rest of the sentence is where it gets interesting.
What each one charges for
| They charge you for | Metabase and Superset | Antvia | Why it matters |
|---|---|---|---|
| Compute and volume | Not charged | Not charged | More data or more refreshes, bigger bill |
| Users | Not charged | Not charged | Every extra reader costs money, forever |
| Queries | Not charged | Not charged | Asking more questions costs more |
| Sources | Not charged | Charged | Each maintained connection meters |
| Metrics | Not charged | Charged | Each published definition meters |
Nothing, in licence terms. The meter is your engineer's time: upgrades, CVE patches, connector breakages, and the week in which a database driver changes underneath you.
The honest split
- A semantic layer, so a wrong number is not expressible
- Query-time enforcement rather than dashboard-level permissions
- Somebody to call when a connector breaks at month end
- Arc and Leap
- No licence fee at all
- Superset has far more chart types and complete SQL freedom
- Metabase is genuinely pleasant for a small team to run
- You keep the source, because it is open source
The difference that is not about money
Superset lets a chart author pick any column and apply any aggregate. That flexibility is the product, and for an analyst it is exactly right. It also means nothing stops somebody summing a column that should have been averaged, or aggregating a snapshot across time. The query runs, a number appears, and it looks precisely like a correct one.
Antvia will not offer that. A measure is something the semantic layer computes, so if it is not defined it cannot be charted. Dimensions stay open, because grouping by the wrong column produces an obviously useless chart and costs you a minute. Aggregating the wrong column produces a plausible number and costs you a decision.
What you are actually taking on
Someone has to run it. Not once, continuously: version upgrades, security patches, the connector that stops working the week your reporting is due, and the migration when a major version lands. If you have an engineer who enjoys that, the maths is genuinely in your favour and you should self-host.
If that person does not exist, the licence saving is real and the total is not.
When open-source BI is the right answer
A small team, an engineer who likes infrastructure, no compliance obligation that needs evidencing, and analysts who want SQL freedom rather than guardrails. Metabase in particular is a pleasure at that size and we would happily recommend it.
Questions we get asked
You use open source too, so what are we paying for? The governance, the access enforcement, the semantic layer, and the decision loop. The engines underneath are open and you keep them. We wrote about exactly where that line sits.
Can we start with Metabase and move later? Yes, and it is a sensible path. Dashboards do not migrate, but the thinking does.