Ask Data — Tableau’s Natural Language Analytics Feature.

Tableau is a business analytics platform that offers interactive dashboards with a variety of data visualizations and multiple data analysis features.

Recently, Tableau released a new feature — Ask Data, which helps users visualize and analyze their data by typing a question in plain language. Instead of building a report from scratch, and adjusting dimensions and filters in Tableau’s traditional drag-and-drop way, users can simply type a question, e.g. “what is the sum of leads per city per month” or “what are the top searched keywords in Dubai Marina”. Tableau reads the question, transforms it into a query and displays the results. Users can then further adjust the query and the visualization of the data. The result can be saved for later use and shared with others.

This new feature enables users to ask a variety of questions and get a grip of the data instantly without any special technical skills and without external help. It also makes it easier to analyze the data without having to export it to a different platform (e.g. Excel), as it is easy to slice and dice, and to create various graphs within the tool itself.

Ask Data does not eliminate the need to build dashboards for specific purposes but it enables users to get easy access to more general data and top-level KPIs.

Key Facts About Ask Data:

  • Ask Data is embedded within Tableau Server and Tableau Online so no additional set up is needed and users will only have access to data sources they are permitted to access.
The available data sources with the respective “Ask Data” tag.
  • As is always the case with Tableau products, the interface is clean, straightforward and user friendly. All the fields of the data source are listed with the respective data types. This helps users get familiar with the available data.
Available dimensions and metrics are shown on the left of the question field and the suggested queries below.
  • Tableau pre-analyzes the fields in the data source and suggests sample questions based on the data types. Once the user starts typing their question even more suggestions appear. This is very helpful for users who might be unsure of how to structure the question and what kind of data they can get from the source.
Once the user starts typing a question, Tableau shows more suggestions.
  • Ask Data gives the option to add synonyms for certain phrases to improve the natural language processing.
  • Even in the Beta version of Ask Data the speed is impressive — I have been testing this on our ‘searches’ data (which is huge, as you can imagine!) and I was amazed at the processing speed.

Important caveats:

  • The data sources should be clean and simple, ideally prepared for Ask Data with all variables and dimensions named properly. Custom variables and hierarchies should be created in advance; data types and geographical roles etc should be pre-setup. More information on how to prepare the data, can be found here.
  • Your users should know what they are looking for and be familiar with the data sources (the structure of the data source, the variable names, etc). They need to use the exact variable names in their questions, and they need to know which fields can and cannot be aggregated etc.

Potential Problems:

The main problem with Ask Data is the fact that some users might not know the right question to ask. Many users don’t know how to structure their questions or don’t know what data they need to query. Letting users ask open-ended questions holds the risk of them getting the wrong data or no data at all.

Hence, the cleanliness of the data source is extremely important. As mentioned above, if the data source is in the right format, Tableau will be able to analyze properly and offer the correct questions and answers.

How to Start:

Log in to Tableau:

“All Projects” page on Tableau Online — the first page users see upon login.

Select “All Data Sources”:

The dropdown menu on the top of the page allows users to select “All Data Sources”.

Search for the data sources optimized to Ask Data (when creating the data sources it is good practice to add a tag, e.g. ‘Ask Data’, to help users identify them) and click on the source you are interested in:

Users can search for the optimized data sources using the filter on the right.

Once a data source is selected, its content will be shown on the left. On the right you will find the question field and the suggested queries

All available dimensions and metrics and the suggested queries.

Start typing questions. Based on the data, there are suggested questions. Once you’ve started typing, more suggestions will appear:

Ask Data suggests a proper question structure as well as further questions.

And this is the result:

Ask Data displays the data in various ways. This is a simple bar chart. The data could be sorted differently (e.g. alphabetically). This needs to be specified in the initial question.

You can change the visualization and see it as a simple table:

The menu in the top-right corner allows users to change the visualization on the go.

You can sort the table by adding “sort by” to your initial question:

Questions can be made very detailed. Tableau suggests the correct structure, based on the keywords and the data source content.

Overall, Ask Data is a great option for sharing data with a larger group of stakeholders and establishing a data-driven culture within your organization. The data sources should be designed properly and basic training should be provided to the future users.

If you liked this article and want to be part of our brilliant team of engineers that produced it, then have a look at our latest vacancies here.

--

--