Use of Q&A with Deployment Pipelines

[Aleaologist]

I'm having some challenges combining the Q&A feature with deployment pipelines, and am struggling to find guidance on how these two features work together.

Here is a summary of the main challenges encountered so far:

1. I can improve the Q&A service within Power BI Desktop by defining synonyms and creating suggested questions, then publishing to the Development workspace.  The Q&A features work correctly for the report in the Development workspace.  However, when I deploy that report to the Test workspace, the synonyms and suggested questions are not deployed with the dataset.  I do not know how to deploy the Q&A improvements as part of a dataset deployment using a deployment pipeline.
2. Q&A allows you to improve the service over time by reviewing questions people have asked and fixing misunderstandings.  However, users will be working with the report version in the Production Workspace, whereas report changes need to be made in the Development version of the dataset, then deployed into Production via the Development and Test workspaces (with suitable change controls).  How am I supposed to do that for Q&A when all of the usage data will be accumulating within the Production report?

Can anybody please advise how these two challenges should be resolved?

[v-tangjie-msft]

Hi @Aleaologist ,

According to this article, the Q&A improvements such as synonyms and suggested questions are stored in the dataset settings, which are not copied during the deployment process. Therefore, you need to manually update the dataset settings in each stage of the pipeline after deployment. Alternatively, you can use the deployment pipelines REST APIs to automate the process of copying the dataset settings from one stage to another.

According to this article, you can use the Q&A Insights feature to monitor and improve the Q&A experience across different environments. The Q&A Insights feature allows you to see the usage data and feedback for each dataset in each environment, as well as compare the Q&A performance across environments. You can also export the Q&A Insights data to Excel or Power BI for further analysis and reporting. To access the Q&A Insights feature, you need to go to the Power Platform admin center, select your host environment, then select Resources > Dynamics 365 apps > Power Platform Pipelines > Q&A Insights.

Best Regards,

Neeko Tang

If this post  helps, then please consider Accept it as the solution  to help the other members find it more quickly. 

[Aleaologist]

Hello @v-tangjie-msft ,

Thank you for taking the time to respond to my post.

I've reviewed your post, and the pages that it links to, and it doesn't quite answer my questions.

I'm aware that deploying a dataset only copies the model and not the data, but I can't see any reference to the Q&A improvement data in the first article that you linked to. Given that synonyms can be edited from the data model view in Power BI Desktop, it feels like this is part of the dataset structure / metadata, just like everything else you edit on that page such as formats, hidden/visible, display folders etc. From what you're saying, it sounds like the synonym data is not considered part of the data model, but the article is not explicit about this.

After reviewing the Deployment Pipelines REST APIs article, I can't see any mention of the ability to copy dataset settings from one pipeline to the other. The list of pipeline tasks available from the API doesn't include any features that aren't available in the Deployment Pipelines user interface.

The article you referenced in your second paragraph refers to a different kind of deployment pipeline, and doesn't mention Q&A.

As a result of the above, I'm still not clear on my first question about why the Q&A improvement data is not included when deploying between stages in a deployment pipeline, given that the MS documentation does not explicitly state this as a limitation, and the synonyms are presented in Power BI Desktop in the model view alongside all the other model data that IS included in a dataset deployment.

I'm also no further with my second question - if I have to do all the Q&A improvement work in the Production report, and if I have to do the majority of that within Power BI Desktop (because the only thing you can do in the workspace is add suggested questions), and then publish that directly back to the Production workspace this completely undermines the intended robustness of using a deployment pipeline in the first place.

It feels like this is a bug or a missing feature, as a result of the Q&A and Deployment Pipeline teams not considering how their features will interact.