Overall the UX is good and the UI is intuitive to use, responsive and performs well. The visualisations are easy to create and a good representation of the data underneath. The integration with dbt is great and allows us to define our semantic layer in code which gives data users a a better experience with access to more accurate and trustworthy data.

We self-host the open-source edition, so I can only speak about the free features. We moved from Looker this year, and the switch went better than we expected. - dbt-native semantic layer. Dimensions, metrics, joins and row-level filters live in our dbt YAML next to the models. A single `lightdash deploy` in CI publishes them. We no longer maintain a second modelling language like LookML. - Content as code. Charts and dashboards are YAML files that we keep in Git, review in pull requests and push with the CLI. This let us migrate dozens of Looker dashboards and saved Looks with a scripted, repeatable pipeline instead of rebuilding them by hand. - Integrations. Lightdash works with our whole stack out of the box: dbt, Snowflake and Postgres as warehouses, our identity provider for SSO, Slack and email for scheduled deliveries, and GitHub Actions for deployment. The REST API works in the free edition, so we automate the rest ourselves. - AI-friendly by design. Because the semantic layer and the content are plain YAML, AI coding agents can read, generate and validate them. We relied on AI agents heavily during the migration, and they still help us build new charts and dashboards. We have not used the built-in AI features, which are part of the paid plans. - Pricing and ROI. The open-source edition has no licence fee, so after leaving Looker we pay only for hosting. The free edition still covers what matters to us: SSO, scheduled deliveries, user attributes for row-level security, and several projects on one instance (internal and customer-facing, US and EU). - Flexibility. When a dbt model was not enough, the SQL Runner and virtual views let us build custom datasets for specific visualisations. - UI, performance and onboarding. The interface is clean and queries feel fast. Onboarding our business users took very little effort: they switched from Looker very quickly and have not reported any loss of comfort. - Support and open source. The documentation answered most of our questions. When something behaved unexpectedly, we read the source code and confirmed the behaviour ourselves, without waiting for a support ticket.

I have been solo running a data team in a startup for a while. What I love about Lightdash is the wonderful control it gives a lean data team to create everything via their CLI. They have a wonderful MCP for senior stakeholders to query, but I can create dashboards (and data apps!) almost entirely without touching the wiziwig editor. Bliss