- Poor data governance: if you opt to write SQL code instead of building everything with Sigma's built in functions, the data lineage that Sigma provides breaks down completely. If your data team is making changes to a data model, you lack the visibility to identify all of the locations where this model is used if said model is used in any workbook SQL query. We have had to disable custom SQL within our instance due to all of these SIGNIFICANT limitations.
- Difficult to Troubleshoot/Maintain Dashboards: if a column or data model is broken, Sigma will replace that column with a hardcoded error, which makes it nearly impossible to resolve. Moreover, there is no unified view to surface errors in Workbooks-- you have to search through every single workbook page and look for red X's on each table/chart to identify what's wrong. Instead of being able to proactively resolve errors, you'll constantly have business stakeholders raising these issues since the tool doesn't surface these to your data team members in any meaningful way.
- Poor DBT integration: DBT integration is almost non-existent, outside of surfacing column descriptions in a few select areas of the UI. These columns descriptions will not persist, however, if you use Sigma's dataset or data model functionality. There is no semantic layer integration.
- Lack of Simple Polish: Grouping columns of a dataset into logical sections is not available, so you're constantly sifting through a large list of columns to find what you need. Sigma metrics are incredibly basic, cannot be reordered/organized, and don't work with their new data model functionality. Schema refreshes are not performed at any set cadence and there is no ability to schedule a daily schema refresh, leading to data modeling displaying stale views of the data without manually refreshing each one individually.
- Limited API: The API has no functionality to manage Sigma metrics, datasets, or data models. It also lacks the ability to trigger a refresh of a workbook, which would allow our team to build out some functionality to address the lack of data governance within the product.
Sigma seems more concerned with flashy marketing and sponsorships and launching half-baked AI functionality than they are in making meaningful improvements to their product. Review collected by and hosted on G2.com.