Introduction#

This article covers best practices for optimizing workbook performance in Tableau Desktop. It provides practical guidance to identify and resolve performance issues, ensuring faster and more efficient visualizations.

General#

As shown in the webinar video, this article is structured into four layers:

• Presentation
• Analytics
• Data
• Platform

Each section focuses on key aspects to help optimize your Tableau workbook’s performance.

Presentation#

Sheet#

• Keep the sheet as simple as possible.
  • Include only the data necessary for analysis. Remove unnecessary elements like parameters, sets, groups, calculated fields, and filters.
• Tableau is for data analysis, not just listing database contents. Use small cross tables for details, but large ones miss the tool’s purpose.
• Minimize the number of marks and avoid complex ones like polygons, donuts, and maps unless essential.
• Limit filters—favor filter actions and guided analysis.

Dashboard#

• A dashboard consists of two types of zones:
  • Data-driven: sheets, filters, parameters, page controls, legends.
  • Non-data-driven: text, images, web content, blank spaces, layout containers.
• Minimize zones; each data-driven zone triggers a query (recommendation: under 10 zones).
• Interacting with dashboard elements (filters, parameters, actions) can trigger events and additional SQL queries.
  • Global filters, filters applied across all sheets, and “Show Relevant Values” filters increase workload. A dashboard with 18 sheets could send 18+ queries—some cached, others not.
  • Use input filters for long lists (e.g., customer names) instead of quick filters to reduce loaded values.
• Avoid large, high-resolution images.
• Improve cache efficiency by setting dashboards and stories to fixed size and using client-side rendering when possible.

Workbook#

• Split large workbooks into separate files and use fewer sheets per dashboard.
• Remove unused sheets and hide those used only in the background.

Analytics#

Data Elements#

• Minimize data type conversions; perform them at the data source when possible.
  • For example, if a date field is stored as a string, convert it at the source. Each conversion in Tableau triggers a query, increasing processing load.

Filters#

• Avoid using “Only Relevant Values”; prefer “All Values in Hierarchy.”
• Use filters cautiously for small data sources.
• User filters and the ISMEMBEROF() function can impact cache sharing.
• Apply filters at the data source level.
• Limit the number of filters and use them efficiently.
• Avoid high-cardinality quick filters (multi-selection or dropdown lists).

Calculations#

• External function calls (R, Python, MatLab) can be slow.
• Integers are faster than dates, and booleans are faster than strings.
• Limit data blending; it requires sequential queries across multiple sources.
  • If possible, create a database view on the server instead.
• MIN and MAX perform better than AVG and ATTR.
• Use Tableau’s native features—groups, sets, bins, custom date fields, combined fields, and aliases—instead of calculations.
• Prefer optimized functions like CONTAINS(), DATEPARSE(), or MAKEDATE() for dates rather than complex calculated fields.
• Avoid COUNTD() when possible; it’s a slow function.
• Use CASE() or groups instead of multiple IF statements.
• ELSEIF is faster than ELSE IF.
• REGEX is highly efficient for string operations.

Working across data sources#

• Do not blend data on high cardinality dimensions.
  • Prefer a join instead.

Data#

Connection Type#

• A native connector is faster than an ODBC connection.
• An embedded data source can be faster than a published data source.
  • However, prefer a published data source for reuse.
• On Tableau Cloud, prefer a direct connection over Tableau Bridge.
  • Check with your security team.

Data Connection#

• Centralize your data for faster access than extracting it from multiple sources.

Connection Models#

• Limit custom SQL queries in live connections, as they may be inefficient.
  • Instead, create a view in the database and connect Tableau to it.
• Avoid parameters in custom SQL. Tableau wraps them in a subquery, which many databases handle poorly.
• Look for unnecessary clauses like ORDER BY. Tableau will re-sort the data once loaded.

Extract#

• Hide unused fields.
• Aggregate data.
• Filter data.
• Use materialized calculations.

Platform#

• Monitor Tableau Server, for example with TabMon.
• Keep your database close to Tableau Server.
• Identify configuration, data, and workbook design issues.
  • For example: a poorly designed workbook with high traffic can consume many server resources (CPU/Memory).
• Update Tableau Desktop and Server to improve performance and fix bugs.
• Upgrade hardware: RAM, CPU, disk (especially for extracts), and fast network.
• Separate VizQL Server nodes from Backgrounder nodes.
• For virtual machines:
  • Virtualization incurs a performance overhead (10-20%) compared to physical infrastructure.
  • Avoid shared virtual machines!

Resources#

• Webinar: Designing efficient workbooks
• Whitepaper: Designing Efficient Production Dashboards
• Streamline your workbook
• Tableau Workbook Performance Checklist