Optimizing Alteryx workflows for large datasets usually comes down to reducing unnecessary data movement and processing early.

A few practices that consistently make a difference:

• Filter and sample early
  Push filters as close to the source as possible. Processing full datasets when only a subset is needed slows everything down.

• Leverage in-database processing
  Use In-DB tools where possible so heavy joins and aggregations happen in the database, not in-memory.

• Optimize joins and data types
  Ensure keys are indexed and data types are consistent. Mismatched types or large string fields can significantly impact performance.

• Minimize tool complexity
  Break complex workflows into smaller, modular components. This improves both performance and maintainability.

• Use caching strategically
  Cache intermediate outputs when iterating, instead of re-running the entire workflow each time.

• Monitor memory usage
  Large datasets can quickly exhaust available memory. Adjust block sizes and avoid unnecessary field expansions.

In practice, the biggest gains come from designing workflows with scale in mind, not optimizing them after they slow down.