Everything you need to install Insights, point it at the right tables, and start applying clustering key recommendations — typically in under 15 minutes, entirely inside your Snowflake account.
Goldilox Insights is a Snowflake Native App. It studies how your queries actually read your tables, then recommends clustering keys that improve partition pruning — so queries scan less data, run faster, and cost less. Because compute is billed by warehouse runtime, a faster query is a cheaper query.
Insights never moves your data. It runs as an app inside your Snowflake account and reads only query metadata and table statistics — never the contents of your tables.
ACCOUNTADMIN, or a role granted the marketplace import privilege).Open the Goldilox Insights listing on the Snowflake Marketplace and click Get. Installing the app gives you a free preview of the Insights dashboard — no recommendations or payment required. There are no external services to deploy and nothing to host.
On first launch, the app's setup screen requests the privileges it needs to analyze your workload. At a high level, Insights needs to:
Grant the privileges through the in-app prompts. Insights does not request access to the contents of your tables.
Choose the tables you want Insights to watch. The best candidates are your largest, most-queried fact tables — the ones where excessive scanning costs the most. Insights highlights high-impact tables with low pruning efficiency so you can prioritize. Your plan determines how many tables you can monitor at once.
Once tables are selected, Insights begins analyzing your query telemetry. It uses trained prediction models to identify the clustering key that would maximize partition pruning for each table's real access pattern. As your workload shifts over time, recommendations adapt.
The dashboard shows, per table:
Because every recommendation carries an estimated saving, you can sort by impact and apply the changes that matter most first.
Insights generates the SQL for each recommendation. Applying a clustering key in Snowflake looks like this:
ALTER TABLE my_db.my_schema.STORE_SALES
CLUSTER BY (SS_SOLD_DATE_SK, SS_STORE_SK);You stay in full control: Insights recommends, you review and apply at your discretion. After Snowflake reclusters the table, new queries benefit from improved pruning automatically — no query rewrites needed.
Clustering is not one-and-done. As query patterns evolve, a key that was optimal can drift. Insights continuously monitors your selected tables and updates recommendations when a better key emerges, so your tables stay tuned without manual re-analysis.
Insights runs entirely within your Snowflake environment. Your data never leaves your account, and only query metadata and telemetry are analyzed — never the contents of your tables. See the Privacy Policy and Terms of Use for details.
Insights needs enough query history on a selected table to model its access pattern. Give it a short window after installation to gather data, and make sure the table has been queried recently.
That's expected — recommendations adapt to your current workload. If your queries have shifted, the optimal clustering key can shift with them.
Estimates are grounded in your actual historical query costs rather than best-case projections, so they tend to be realistic rather than optimistic.
Questions or issues? Email support@goldilox.com or request a demo.
Ready to start? Install Insights and preview the dashboard for free.
Try Free on Snowflake Marketplace