Video |

Tableau | Data visualisation | Analytics

Tableau 2020.2: Data roles in Ask Data

Data roles in Ask Data let you bank synonyms once and reuse them across data sources.

Tim Ngwena •6 May 2020 •1 min read •4 min watch •1.2K

Part ofWhat's new in Tableau 2020.2

Watch on YouTube

Data roles let you add synonyms and edit field descriptions in Ask Data so users can phrase natural-language questions more flexibly
You can add synonyms for both a field name (e.g. 'demographic' for segment) and its individual domain values (e.g. 'business' for corporate)
Always hit enter after typing a synonym so Ask Data stores it correctly
Publishing a data role to a project folder makes those synonym assignments reusable, and you can apply different saved roles to swap assignments in and out

0:00In Tableau 2020.2, Tableau have added

0:02several features to the

0:03ask data capability, most notably data

0:06roles. Data roles allow you to do some sort

0:09of preliminary

0:10work, add synonyms, you know, change

0:13descriptions to specific fields in ask data

0:16and then apply

0:17those as a published role so other data

0:19sources can then use that information

0:22themselves, okay.

0:23So let me show you how that works. I'm

0:25going to head over here to explore. I'm

0:27going to hit the

0:28top level drop down and I'm going to go to

0:30one of my data sources. It's called Super

0:33store data

0:33source, how original that is, okay. And you

0:36can see here that ask data is actually

0:38enabled on this

0:39particular data source. Now if I go over to

0:42segment here, you can see that I get this

0:44sort of pencil

0:44icon when I hover over it and I get a

0:46little preview of what's actually in the

0:48field. I've got

0:49three domain values, consumer, corporate

0:51and home office. It's a string field and it

0:54has three

0:54unique values, the ones I've just sort of

0:56highlighted. And then if I click on this

0:58edit

0:58icon, I actually get this interface that

1:01allows me to add a data role, okay. So what

1:04does this

1:05actually mean? Well, if I look at this

1:07field, I want it to be easy for users to be

1:10able to type

1:11and ask questions in Tableau and ask data.

1:14So what I need to be able to do is choose

1:17words here,

1:17for example, that could be synonymous with

1:20a segment. So I might just say a

1:22demographic,

1:22so for example, if I just click inside of

1:25here and I just type in demographic and I

1:28hit enter,

1:29you'll see that it's added that there as

1:31well. I'll also expand these three values

1:34and you'll

1:34see that it's loading these fields but

1:37sometimes it's a little bit buggy. This is

1:40a beta version,

1:40so now you can see it's loaded them up. So

1:43now here I can just say consumer, maybe

1:46someone might

1:47type it. So let me just hit consume and see

1:49what happens. I'll call this business as

1:51well because

1:52sometimes you might call your corporate

1:54customers your business customers, right.

1:57So let's just,

1:58let's make sure we hit enter as we type

2:00that. So I'm actually clicking out. What I

2:02need to do is

2:03hit enter and then you'll see that it

2:05actually stores that particular item there.

2:07And then I'll

2:08just say add another synonym for this which

2:10is home. So I've got three synonyms there

2:13and one

2:13for the main field itself, the segment. If

2:16I drop that down, you can see that here.

2:18Now what I can

2:20do is I can actually add this as a saved

2:22data role. So let me publish this data role

2:25up. I'm

2:26going to publish it to my 2020.2 release

2:29folder. I need to give it a name. So I'll

2:31call this segment

2:32roles. Okay. You need to call things in a

2:35logical sense. I'll zoom back out a bit so

2:37we can see a

2:38bit more of this content and then I'm going

2:40to hit publish. Okay. And when we publish

2:43this out,

2:44you'll see that this is now actually

2:47published and it now acquires that segment

2:50role just there.

2:51Now when I click on that again, you now see

2:53a different interface because you can see

2:56it says

2:56that I've applied this data on myself, but

2:59more importantly, I can actually apply a

3:02different role.

3:02So you can see here that I created one a

3:04little bit earlier in another project and I

3:07can actually

3:07click on that hit apply and it will pick up

3:10those assignments. So it removes the

3:13demographics in an

3:14M that I have there and it adds those three

3:16back in because I actually previously typed

3:19them in.

3:20But if I just go back and I apply a

3:21different role again, one of the new ones I

3:24created hit apply,

3:26demographics comes back and the three

3:28unique values get the new consume business

3:31and home

3:32options. Okay. So that's pretty much the

3:34feature in a nutshell. That's the feature

3:37sort of mostly

3:37covered. Check out the other videos on the

3:39new ask data capabilities that are

3:41available

3:42for more information. Okay, that's it

3:44pretty much. If you've enjoyed the video,

3:46hit the like button,

3:47hit subscribe. Otherwise, I'll catch you in

3:49the next video. Thanks for watching.

Rollup

Data roles in Ask Data let you add synonyms and edit field descriptions once, then publish and reuse that work across data sources so users can phrase natural-language questions more flexibly.

Demoed on a published data source with Ask Data enabled, using a segment field with domain values like consumer, corporate and home office.

The Breakdown

Open the field edit interface 0:42

Hover over a field in Ask Data's field list to see a pencil icon and a preview of its type and domain values, then click the icon to open the edit interface where you add a data role.
Add synonyms for the field name 1:02

Type alternative words a user might type instead of the actual field name (e.g. 'demographic' for a segment field) so Ask Data recognises more phrasings of the same question.
Add synonyms for individual domain values 1:40

Expand the field's values and add synonyms for specific values too, not just the field name itself (e.g. 'business' for 'corporate', 'home' for 'home office'), since users may refer to values by different terms.
Always hit enter after typing a synonym 1:58

Clicking away without pressing enter won't store the synonym; you must hit enter each time so the entry actually saves before you move on.
Publish the synonym work as a data role 2:20

Once your synonyms and descriptions are set up, publish them as a named data role to a project folder, which turns this one-off effort into a reusable asset with a logical name.
Apply saved roles to other data sources 2:41

After publishing, the field shows which role is currently applied; you can swap in a different previously-saved role and hit apply to instantly pick up its synonym assignments instead.
Switching roles swaps assignments in and out 3:02

Applying a different role removes the current synonyms and replaces them with whatever was saved in that other role, letting you reuse different synonym sets across multiple data sources without redoing the work.

Worth Knowing

This was a beta version at the time, and the interface was noted as a little buggy when loading domain values.
Forgetting to hit enter after typing a synonym means it won't be stored.

Use It When

Reach for this when multiple data sources share similar fields and you want users across your organisation to get consistent, flexible natural-language search in Ask Data without re-typing synonyms for every source.

How this Rollup was made provenance & method

A Rollup is drafted by AI from the video's transcript, then reviewed and edited by Tim. Everything used to produce this one is listed below — the model, the exact prompt, and the source video — so the process is transparent and reproducible.

Transcription: On-device — NVIDIA Parakeet v3 for recent videos, OpenAI Whisper large-v3 for earlier ones. The transcript never leaves the machine or gets published.
Drafting: Claude Sonnet 5 in the cloud, from that transcript.
Prompt: The exact Rollup prompt (v2) — the full system prompt, unedited.
Source video: Watch on YouTube
Drafted: 5 July 2026 at 09:38
Reviewed & edited: 5 July 2026 at 09:41 · by Tim Ngwena

Model + prompt + video is everything you'd need to recreate a Rollup like this yourself. The one thing we don't share is the transcript.

Rights. The video and its transcript are the property of TN Media Ltd. Unauthorised use or download is prohibited. © TN Media Ltd.