
Semantic layer, Lightdash AI, and Lightdash MCP are the main reasons I’m using Lightdash. They ship very frequent updates and keep adding new features, which I really appreciate. Unlimited seats is another big plus. Overall, nearly everything is code-based, which helps a lot with AI workflows. Review collected by and hosted on G2.com.
Overall, there’s no major issue; it’s more a collection of small friction points here and there. A few that come to mind as of today (November 2026): having to re-run a query after making changes to a chart; in Explore, when searching for a dimension/metric, the search only checks and matches the object name and not the descriptions; in Explores, the field-description pop-up windows are small and require an extra click to “Read full description” when it's a bit long; there’s no query cost estimate in the Explore SQL panel, despite working with BigQuery; and the “Custom Dimension” window feels annoyingly small, even when “expanded”. That said, you could argue that if you let Lightdash AI do the work for you, you won’t be bothered by these few bits. Review collected by and hosted on G2.com.