1. Use Extracts Instead of Live Connections
Why it matters: Live connections rely on the database for every query, and performance can take a hit—especially if network traffic is high or the database isn’t optimized.
How to do it:
- In Tableau Desktop, go to Data > [Data Source] > Extract Data > Use Extract
- Alternatively, from the Data Source tab at the top right select Extract instead of Live
- Use filters to reduce the data to what you need
- Save the .hyper file locally or publish it to Server/Cloud
- Set up a refresh schedule if the data updates regularly
Pro tip from the deck:
- Use incremental refresh for large datasets
- Hide unused fields and aggregate where possible to reduce extract size
2. Hide Unused Fields
Why it matters: Reducing the number of fields and rows Tableau processes improves performance and keeps your data model clean.
How to do it:
- To the right of the search bar above your fields click the drop down arrow, select hide unused fields
3. Pre-compute Calculations in the Database
Why it matters: Calculations—especially row-level and LODs—can add query load. Offloading these to your database saves Tableau from recalculating them on the fly.
How to do it:
- Identify repeated, complex Tableau calculations
- Ask your data engineer to create views or calculated fields in the database
- Replace Tableau formulas with those fields
- Avoid Custom SQL when possible. If used, make it simple and specific
- Custom SQL prevents join culling and can result in subqueries that slow things down
4. Remove Unused Worksheets, Data Sources, and Device Layouts
Why it matters: Anything unused still gets stored and loaded. This adds workbook bloat, increases memory usage, and affects load time.
How to do it:
- Audit your workbook: delete testing sheets and orphan data sources
- Go to Dashboard > Device Preview and remove unused device layouts
- Run Workbook Optimizer (Server > Workbook Optimizer) to catch low-hanging cleanup opportunities
5. Reduce the Number of Views Per Dashboard
Why it matters: Each view is its own query. Too many views = too many queries = slower load.
How to do it:
- Focus on key insights—remove “nice-to-have” visuals
- Use navigation buttons to guide users to secondary dashboards
6. Be Strategic with Filters
Why it matters: Filters are deceptively heavy. Each one generates a query. Filters like “Only Relevant Values” multiply that query load.
How to do it:
- Use fewer filters and avoid high-cardinality ones where possible
- Use parameters or dashboard actions instead of traditional filters
- Always enable the Apply button for multi-select filters
- Prefer Data Source Filters or Extract Filters over context filters
7. Benchmark Performance with a Performance Recording
Why it matters: You won’t know what’s slowing your dashboard until you look under the hood. Performance Recording shows query times, rendering delays, and lets you pinpoint the slowest steps.
How to do it (Desktop):
- Help > Settings and Performance > Start Performance Recording
- Use your dashboard as a typical user would
- Then stop the recording and Tableau opens a workbook showing:
- Query time
- Render time
- Layout processing
How to do it (Server):
- Add ?:record_performance=yes at the end of the dashboard URL, before :iid= Example: https://yourserver/views/Dashboard?:record_performance=yes&:iid=1
- To avoid cache skewing results:
- Create and save a blank sheet
- Reopen the workbook on that blank sheet before recording
8. Use the Optimize Workbook Feature
Why it matters: This feature automatically flags performance issues and helps you clean up your workbook in a structured way.
How to do it:
- Server > Run Optimizer
- Tableau will categorize items as:
- Take Action (safe to fix immediately)
- Needs Review (use judgment)
- Passed (best practices followed)
- It’s a fast way to catch unused fields, calc issues, and stray elements
9. Understand the Bigger Picture
Why it matters: Performance isn’t just about what’s in Tableau—it’s shaped by the full pipeline: the underlying infrastructure, data quality, and how you’ve connected things.
- Prefer native drivers over ODBC/JDBC when possible
- Understand the impact of joins vs relationships vs blending
- Pay attention to number of marks (displayed in bottom left of Tableau Desktop)
- Use published data sources on Server/Cloud for shared, consistent, and optimized access
Final Thoughts
Improving Tableau performance is about knowing what to simplify and where to shift the load. If you're short on time, my go-to sequence is:
- Switch to an Extract
- Use Optimize Workbook
- Hide unused fields + filters