Course: Tableau Training
Topic: Understanding Performance Recording
Learning Objective: Understand Tableau Performance Recording, learn how to capture workbook performance data, interpret recorded events, identify performance bottlenecks, and apply appropriate optimization techniques.
Register for a Tableau Course Demo
Performance Recording is a built-in Tableau diagnostic feature that captures timing information about important activities performed while a user interacts with a workbook. It helps developers and analysts understand which operations consume the most time and where optimization may be required.
When a Tableau dashboard loads slowly, the cause might be a database query, a data-source connection, a complex calculation, dashboard layout processing, extract generation, or another operation. Performance Recording helps investigate these possibilities using recorded evidence rather than guesswork.
Suppose a sales dashboard takes 18 seconds to load. You suspect that the dashboard contains too many charts. However, a Performance Recording shows that most of the delay comes from a slow database query. Instead of redesigning every chart, you can investigate the query, data model, and source database first.
Tableau Performance Recording captures selected events that occur as a workbook is used. The events available depend on the Tableau product, version, and environment.
| Event | Meaning | Possible Performance Issue |
|---|---|---|
| Connecting to Data Source | Time spent establishing a connection to a data source. | Network delays, authentication issues, or a slow database server. |
| Compiling Query | Time spent preparing a query for execution. | Complex workbook logic, filters, or calculations. |
| Executing Query | Time spent executing a query against a data source or extract. | Large data volumes, expensive queries, or an inefficient source model. |
| Computing Layouts | Time spent calculating the layout of a view or dashboard. | Complex dashboard layouts, many objects, or demanding visualizations. |
| Generating Extract | Time spent creating an extract. | Large extracts or unnecessary fields and records. |
| Geocoding | Time associated with processing geographic information. | Large amounts of geographic data or unnecessary geographic detail. |
| Blending Data | Time associated with combining data from different sources through data blending. | Large datasets or inefficient data-blending operations. |
| Server Rendering | Server-side rendering activity, where supported by the recording environment. | Server workload, rendering demand, or resource constraints. |
Important: A recorded event identifies where time was spent; it does not always identify the exact root cause. Additional investigation may be required.
Performance Recording follows a simple process: start recording, perform the actions you want to investigate, stop recording, and analyze the resulting performance workbook.
Open Tableau Workbook
|
v
Start Performance Recording
|
v
Perform User Actions
(Open Dashboard, Change Filters, Select Marks)
|
v
Stop Performance Recording
|
v
Open Performance Recording Workbook
|
v
Analyze Timeline, Events and Queries
|
v
Identify Bottlenecks
|
v
Optimize and Test AgainFollow these steps to create a performance recording for a workbook in Tableau Desktop.
Help
→ Settings and Performance
→ Start Performance Recording
Perform workbook actions
Help
→ Settings and Performance
→ Stop Performance RecordingTip: Record a specific, repeatable user workflow. If the issue occurs when a dashboard opens, focus on opening that dashboard. If a filter is slow, record the filter interaction separately when possible.
Official documentation: Record and Analyze Workbook Performance in Tableau Desktop.
After you stop recording, Tableau generates a workbook that summarizes the captured events. The recording workbook commonly includes two main dashboards: Performance Summary and Detailed Views.
The Performance Summary dashboard provides a high-level view of the recorded activities. It helps identify which events took the most time and when they occurred.
Its main sections include:
The Detailed Views dashboard provides more technical information about activities performed during a request. It is intended for deeper investigation and is available when the performance recording workbook is opened in Tableau Desktop.
It can include views such as:
| Dashboard | Main Purpose | Best Used For |
|---|---|---|
| Performance Summary | Provides an overview of recorded performance events. | Quickly locating the longest-running event categories. |
| Detailed Views | Provides deeper information about activity timing and hierarchy. | Investigating the internal steps involved in processing a request. |
The Timeline view shows recorded activities in the order in which they occurred. The horizontal axis represents elapsed time, while the event details identify the workbook context and the type of activity.
Suppose the timeline shows that a query took 8 seconds, followed by layout computation taking 2 seconds. This suggests that query execution deserves investigation first. However, you should also check whether the events overlap, repeat, or depend on other activities before concluding how much time each contributes to the overall user wait.
The Events view helps prioritize performance issues by displaying recorded event types and their durations. Longer-duration events often provide a useful starting point for troubleshooting.
A recording might show these illustrative durations:
| Event | Illustrative Duration | Initial Investigation |
|---|---|---|
| Executing Query | 8 seconds | Review query complexity and data-source performance. |
| Computing Layouts | 3 seconds | Review dashboard structure and visual complexity. |
| Connecting to Data Source | 2 seconds | Review connection and network conditions. |
| Other Events | 1 second | Investigate if significant to the overall delay. |
Note: These figures are examples for learning, not actual Tableau benchmark results. Do not simply add every event duration together, because some activities may overlap or be nested.
The Query section can show the text associated with a selected query event. Depending on the data connection and Tableau environment, this may be SQL, XML, or other query-related information.
SELECT Region, SUM(Sales)
FROM Orders
WHERE OrderDate >= '2026-01-01'
GROUP BY RegionThis illustrative SQL query retrieves sales totals by region for orders from a specified date onward.
If a similar query is slow, a database administrator might investigate the execution plan, data volume, table structure, indexes where appropriate, and filtering strategy.
Important: Query details may not be available for every event or connection. In some Tableau Cloud scenarios, especially with published data sources and certain versions, the recorded event naming and query details may differ.
The Depth view displays activities in a hierarchy. A single user request, such as opening a view or changing a filter, can trigger several related activities.
Higher-level activities may contain lower-level child activities. Reviewing this structure helps explain what Tableau did while processing a request.
Example: A filter interaction may lead to a query being prepared, data being retrieved, and a visualization being updated.
Exclusive CPU represents CPU time attributed directly to an activity, excluding CPU time attributed to its child activities. It helps identify activities that consume processing resources themselves.
Inclusive CPU includes the CPU time of an activity and its child activities. It helps show the total CPU work associated with a broader activity.
Elapsed time measures the duration between the beginning and end of an activity. It is not necessarily the same as CPU time because elapsed time can include waiting for database responses, network operations, or other resources.
| Metric | What It Measures | Useful Question |
|---|---|---|
| Exclusive CPU | CPU time attributed directly to an activity. | Which activity itself is using processing time? |
| Inclusive CPU | CPU time for an activity including its child activities. | How much CPU work is associated with this activity and its descendants? |
| Elapsed Time | Wall-clock duration of an activity. | How long did the activity take from start to finish? |
Best practice: Use these metrics together. A long elapsed time with relatively low CPU time may indicate waiting rather than intensive local processing, but further investigation is needed to determine the actual cause.
A performance bottleneck is an operation or resource constraint that significantly contributes to a delay in a workbook or dashboard.
| Observation | Possible Explanation | Recommended Action |
|---|---|---|
| Long connection time | Network, authentication, or database connection delays. | Check connectivity, authentication, and source availability. |
| Long query compilation | Complex query generation or workbook logic. | Review complex calculations, filters, and workbook structure. |
| Long query execution | Slow data-source processing or expensive query operations. | Review query plans, source data, joins, and data volume. |
| Long layout computation | Complex dashboard or view layout. | Simplify dashboard objects, labels, and visualization design. |
| Long extract generation | Large extracts or excessive included data. | Reduce unnecessary fields and records when business requirements allow. |
| Long geocoding activity | Large amounts of geographic information. | Reduce unnecessary geographic records or detail. |
| Long data-blending activity | Expensive data blending operations. | Review data volume, filtering, and whether another data model is appropriate. |
Performance Recording helps narrow down the problem. Some causes, such as server saturation or network congestion, may require additional server, database, or infrastructure monitoring.
Tableau Server supports performance recording for views when the feature is enabled for the relevant site. The setup differs from the Tableau Desktop workflow and may require administrator permissions.
Important: Tableau Server setup is different from Tableau Desktop. Use the official instructions for your Server version and operating system rather than copying a URL pattern without checking the current session URL and documentation.
Official documentation: Create a Performance Recording on Tableau Server.
A sales dashboard contains multiple charts, several filters, and a live connection to a database. Users report that the dashboard loads slowly and that changing the Region filter causes a delay.
| Metric | Before | After |
|---|---|---|
| Dashboard load time | Record measured value | Record measured value |
| Region filter response | Record measured value | Record measured value |
| Longest event | Identify from recording | Identify from new recording |
| Data correctness | Validate totals and filters | Repeat validation |
Expected learning outcome: You should be able to explain which event contributed to the delay, what you changed, and whether the change improved performance. Do not claim an improvement until you have measured it.
| Mistake | Why It Is a Problem | Better Approach |
|---|---|---|
| Recording unrelated interactions | Makes it harder to isolate the source of a delay. | Record one clearly defined workflow at a time. |
| Optimizing without measuring | The change may not address the actual bottleneck. | Use the recording to guide the investigation. |
| Assuming all slow queries are Tableau problems | The underlying database may be the main source of delay. | Review query information and source performance. |
| Adding all event durations together | Some activities may overlap or be nested. | Interpret durations in context using the timeline and detailed views. |
| Ignoring data correctness | Performance changes may unintentionally alter results. | Validate calculations, filters, and totals after each change. |
| Expecting identical results every time | Caching, network conditions, server load, and source activity may vary. | Run repeatable tests and compare multiple observations when appropriate. |
Answer: Performance Recording is a diagnostic feature that captures timing information about workbook activities so users can investigate slow dashboards and identify possible performance bottlenecks.
Answer: It helps determine where time is being spent during workbook interactions and supports evidence-based performance optimization.
Answer: Select Help > Settings and Performance > Start Performance Recording, perform the actions you want to analyze, and then select Stop Performance Recording from the same menu.
Answer: It is a high-level dashboard in the generated recording workbook that helps analyze the timeline, event durations, and query information.
Answer: It shows recorded events in chronological order and helps users understand when activities occurred and how they relate to the recorded workflow.
Answer: It presents recorded event categories and their durations, helping users prioritize the operations that may deserve investigation first.
Answer: Query compilation is the time spent preparing a query, while query execution is the time spent running the query against the data source or extract.
Answer: Elapsed time is the wall-clock duration of an activity. CPU time measures processing time attributed to CPU work. An operation can have long elapsed time while waiting for another resource.
Answer: It provides deeper diagnostic information about activity hierarchies, CPU time, and elapsed time for advanced performance investigation.
Answer: No. It helps identify where time is spent, but additional investigation may be necessary to determine whether the cause is the database, network, workbook design, server resources, or another factor.
Answer: When query information is available, developers can review the generated query and work with database administrators to investigate execution plans, data structures, and query performance.
Answer: Yes. Tableau Server supports performance recording for views when the feature is enabled and configured appropriately. The setup differs from Tableau Desktop.
Answer: Repeating the same workflow makes it easier to compare performance before and after the change and evaluate whether the optimization helped.
Answer: Not necessarily. A long event is an important clue, but events can be nested, dependent on other activities, or affected by external resources. Further investigation may be needed.
Create a short report with the following headings: Problem, Recorded Workflow, Main Event, Suspected Cause, Optimization Applied, Before-and-After Measurements, and Validation Result.
Understanding Performance Recording is an important step toward building efficient Tableau dashboards. It helps analysts and developers observe the time spent on queries, connections, layout calculations, extract generation, and other recorded operations. By analyzing the Timeline, Events, Query information, and Detailed Views, users can prioritize performance problems and investigate their causes systematically.
The recommended workflow is to reproduce the problem, record the relevant actions, identify the bottleneck, make one targeted improvement, and record the same workflow again. Always confirm that the optimization improves performance without changing the accuracy of the results.
Explore the course: JustAcademy Tableau Training
Register for a demo: Register for a Tableau Course Demo
