Antvia Intelligence
vs Looker
Looker is the closest thing to a philosophical ally in this category. It believes what we believe about defining metrics once, and it is very good at it.
Who arrives here
You already understand why a semantic layer matters, probably because you have used LookML or been sold it. So the question is not whether metrics should be modelled. It is what else you get, and what it costs.
Where Looker stops
Looker takes you to a trustworthy number and stops, which is where every BI tool stops. Somebody then calls a meeting, a change gets made, and six months later nobody can say whether it worked, because the number moved for four reasons at once and none of them were written down.
What each one charges for
| They charge you for | Looker | Antvia | Why it matters |
|---|---|---|---|
| Compute and volume | Not charged | Not charged | More data or more refreshes, bigger bill |
| Users | 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 |
The honest split
- Arc and Leap: Looker models metrics but stops before measuring decisions
- Unlimited readers
- Materially lower total cost for a mid-market business
- LookML is mature, expressive and battle tested over many years
- Deep Google Cloud integration
- Embedded analytics at real scale
- A genuine developer ecosystem around the modelling layer
LookML against Antvia data models
LookML is mature, expressive and has years of production behind it. Ours is younger and deliberately narrower: it is aimed at somebody who will never write a modelling language, which means it does less on purpose.
If you have an analytics engineer who will invest in LookML, you will get more out of it than out of us on that axis. If you do not, LookML is a language nobody in the building writes, and the semantic layer stays empty.
The part that is genuinely different
Arc records a proposal against the metric it is meant to move and the units it applies to, and refuses one that cannot name both. Leap measures what happened afterwards, against the prediction and against the units the change never touched, and says so plainly when two overlapping changes make the effect impossible to separate.
Looker has never claimed to do this. Nobody in the category has.