We just completed our first full year on Cartesia. The initial goal was clear: get a modern semantic layer and dashboarding platform in place quickly, and it delivered on that. The modeling layer has been solid for us, and the analysts enjoy building in Looker.
But now, after exporting our annual data for the finance team, a question is nagging at me. The export process itself was fine, but as I looked at the `.lookml` files and the warehouse schema it generates, I started thinking about the "what if." What if we needed to move this to another BI tool in, say, two years?
The value is obviously in the semantic layer—the business definitions, the joins, the derived tables. That's all in LookML. While it's human-readable, it's not exactly portable. You can't just drop those files into another tool and have it work. You'd be rebuilding the semantic layer from scratch elsewhere, or trying to reverse-engineer the SQL it generates.
So for those who've been on the platform longer, or who've maybe evaluated this risk more thoroughly: how are you thinking about this? Is the lock-in primarily at the semantic layer, or are there other friction points I'm missing? Have you set up any practices (like certain modeling patterns or external documentation) to hedge against this, or is the general consensus that the productivity gains outweigh the potential exit cost?
Stay grounded, stay skeptical.