Case Study: LogRocket maintains 99.99% accuracy on key tables with Metaplane

Key results

The challenge

LogRocket's two-person data team runs a stack of BigQuery, dbt, Fivetran, Hightouch, Metabase, and Metaplane supporting product, finance, sales, marketing, and customer success. Existing dbt tests and open-source runtime alerts did not catch data drift on a business-critical Stripe source, and manually tracing which Metabase dashboards were affected by model changes was slow. Presenting slightly different revenue numbers each month risked losing the finance team's trust.

The solution

LogRocket deployed more than 50 Metaplane monitors on business-critical sources such as Salesforce and Stripe, tracking row counts, freshness, and monthly revenue sums with group-by monitors, manual thresholds, and sensitivity settings. Column-level lineage and regression testing via the Metaplane GitHub app forecast the downstream impact of dbt changes before merging.

The results, in context

LogRocket has maintained values in key tables to within 0.01%, meeting a 99.99% accuracy SLA. Metaplane was the first to alert on a Fivetran connector outage loading Stripe data, and downstream lineage identified every affected BigQuery table and Metabase dashboard needing cleanup.

Products used