We're evaluating a replacement for our current Tableau Server deployment, which is becoming cost-prohibitive and difficult to maintain for our 200 retail operations users. The primary need is a robust, centralized dashboard for daily KPI tracking (sales, inventory, labor hours) across 500+ stores. Performance with sub-second load times on filtered views is non-negotiable.
Our data stack is built around a Snowflake cloud data warehouse, with transformation handled by dbt. The critical requirement is seamless, low-latency connectivity to Snowflake without constant manual intervention. We've narrowed the field to Domo and Sisense, but the architectural approaches differ significantly.
Domo pushes its own "Cloud Amplifier" for Snowflake, which essentially creates a synchronized copy of your data in Domo's proprietary storage. This can improve dashboard performance but introduces data staleness concerns and duplicates storage costs. Their connector management is more turnkey.
Sisense, particularly with its "Elasticube" technology, traditionally favored a pre-aggregated data model built into its own engine. However, their newer "Compose SDK" and "Live Connect" options claim to enable direct querying against Snowflake, which would be our preferred route to avoid data lag. The documentation on performance with 200 concurrent users in this live query mode is less clear.
From an observability standpoint (my own bias noted), neither offers the granular, code-level tracing we instrument with Datadog for our applications. However, for this BI use case, we need to understand the operational overhead: which platform provides better transparency into query performance, user load, and data refresh failures? Initial testing shows Sisense exposes more detailed query logs for Snowflake, while Domo's operational metrics are more focused on their own ecosystem health.
The decision seems to hinge on this data architecture trade-off: Domo's potentially faster but duplicated data layer versus Sisense's potential for direct, live queries at the risk of straining our Snowflake warehouse with 200 concurrent dashboard users. Has anyone implemented either at this scale specifically for retail ops with a live Snowflake connection?
null