Slow loading Tableau dashboards are one of the biggest issues our clients ask us to solve. You should be aiming to make your Tableau dashboard load in seconds, not minutes. More to the point, a 6 second load time for all dashboards.
When creating dashboards, it’s critical to consider both the end-user as well as the data when considering the design.
Not only should the users’ needs be considered in the context of the data, but the shape of the data used should also be considered in the context of the user’s needs.
How long should your Tableau dashboards take to load?
- Less than 6 seconds: Best practice
- Less than 20 seconds: Acceptable
- Greater than 20 seconds: Poor
- Greater than 60 seconds: Appalling
Our top 5 Tips to make your Tableau Dashboard load in seconds
These quick tips should provide performance improvements across a wide range of existing dashboards.
Here are our 5 top tips for building dashboards with lighting fast load times.
1. Pre-filtered workbooks
If you have a large dashboard with tens of thousands of rows, then default all filters to Null when loading large dashboards. Dashboard interactors should be presented a dashboard quickly and then given the choice of filtering and drilling into the data.
For large dashboards, consider filtering the default view on load to show no records or only a summary view of the data. Once loaded, give the user a clear indication that they can drill into the dashboard via any of the filters.
2. Experiment with connecting to data
Depending on the data sources you access, you may have a variety of ways to connect to it. Experiment with the various connection types to find the optimal.
When connecting to your data, carry out experiments that compare the performance of certain connection methods and data structures. For example, compare live connections, direct extracts, and tailored custom data extracts. Beyond data connection, experiment with the way Tableau filters, parameters, and calculated fields behave with each of the different sources.
3. Extract Size
Reduce extract size wherever possible.
Tableau has a theoretical limit to the size of extracts, and while this is in the billions, extracted data source over a few hundred million rows will experience speed issues. When considering the size of the extract, consider the number of rows that are needed for the visualisation layer.
Consider these actions:
- Filtering the data prior to extract removes any rows outside of the context of the dashboard
- Pre-aggregated extracts utilises the Aggregation function
- Limit the number of rows extracted prior to the extract running
4. Hide unused fields
Connecting to very wide data sources is a benefit of Tableau, however, to ensure performance is maintained, hide any columns that are not used in the visual layer. This ensures that Tableau doesn’t waste time indexing and optimising these unnecessary fields.
Hide all unnecessary fields, and when creating the extract, click the “Hide all unused fields” checkbox. This reduces the size of the extract down to only the number used columns multiplied by the number of rows.
5. Get close to your audience
Communicate with the users regularly.
Involve the users as regularly as possible to find out why they use each dashboard. Discover what they want to see, what they don’t like, what their biggest pain point is, etc.
Use these interactions as an opportunity to align expectations with what can be achieved with Tableau. Compromise on functionality where the speed of delivery is at risk.
Bonus 6. Get help from an expert
Tableau can throw up some really complex barriers that can take days to solve. Stackoverflow is one of the best resources to discover how to overcome some of the more complex barriers that Tableau has to offer. But it can take time to find the problem and then implement the solution.
For the more mission critical dashboards, it will be much quicker to get in a Tableau expert to get the dashboard delivered on time.
Once complete, you can investigate how the challenge was solved.
These quick tips to make your Tableau dashboard load in seconds should provide performance improvements across a wide range of existing dashboards.
Testing and experimentation should form part of the creation of any Tableau dashboard to ensure the end-user is being provided high quality dashboards.
When creating dashboards, it’s critical to consider both the end-user as well as the data when considering the design. Not only should the users’ needs be considered in the context of the data, but the shape of the data used should also be considered in the context of the user’s needs.
If the user requires a view of quarterly sales performance for the customer with the ability to drill into monthly data on an individual, consideration should be made of the size of the data needed to support both views.