Data Portal - My Listings

CVS

Overview

Data Portal is where producers and managers upload and store data. They use this data to create and computes predictive models, but the predictive model is only as good as the data that was used. There is constant conversation between the data producers and the data managers and users.

It is also the repository where users pull data. Before the Data Portal there was no centralized place for both of these user groups (producers and managers) to use and view data.

Where we started

An MVP was quickly released in order to gather user feedback. There are different forms of metadata collected for each type of data. The manager of the data is the one who uploads and owns the individual types of data. They are also the one that edits or accepts suggested edits on the data to create a higher metadata score.

The Problem

Managers own more data points (schema, table, or view) than originally anticipated which was making it hard to find and search for particular assets.

MVP Manage page

MVP Manage detail page

Card Sort

By interviewing users and doing a card sort I was able to determine the most important top level information that needs to be displayed as well as some nice-to-have secondary information.

Top level information

  • type (schema, table, view, report)

  • name

  • status

  • role (owner or steward)

  • modified date

I was also able to pull from the card sort what filters would be most helpful when searching for a data point.

 

User Interviews

Prominent information pulled from user interviews

  • A need for the “name” and “title” of the data point. While the “title” might be a non human readable text it is still how many users are choosing to process the data as there can be duplicate of the human readable “name”.

  • Users who manage just a specific child asset (table or view) don't necessarily understand that there is a hierarchy system in place and that their point lives within a parent asset (schema).

    • This determined, for now, this was not the place to show hierarchy or a nested view of the data. The need was to make it more of a repository to search upon and making each individual asset its own free entity.

  • Reluctance to add data in the portal due to metadata scores being low, even if it was easier edit the data in the portal to increase the metadata score.

    • With this understanding of the users mentality I was able to add a list and ulist feature so that the pressure to have everything in order before registering the data was counteracted.

    • I also took this as an opportunity to display on the detail page what areas of the metadata could use some looking at to bring the score up to par.

Detail page - detail view of list and unlist, metadata check list card

Detail page - detail view of list and unlist as a child data point in a parent schema

 

My Listings page detail - Pending status

Journey Mapping

From the user interview I was able to create a journey map that showed the thought process of when a user is registering data to the point of searching for a specific data point. This helped to identify gaps in the experience.

Users were going in multiple times to register data because of the time lag to publish the data in the experience.

This gap in the experience made it clear that I needed to put in an informational design element to tell the user that they had already gone through the registration process but the crawl had not fully brought the data over yet. This is why I implemented thee “pending” status.

 

Testing Designs

After I created the designs and took them through some user testing/validation, it became clear that while the list and unlist feature would be greatly utilized it created some logical gaps due the parent child relationship of the data. With this in mind I set up a matrix and created some logic to counteract those gaps.

  • Schema has to be listed if one table/view is listed

  • Schema can be listed with no tables/views listed

  • Only the owner of the schema can list or unlist full schemas

  • If you own the table/view you can unlist it

  • If you only own table/view you can only list it if the schema is listed.

  • With these rules in place we added different pop up messaging depending on the situation.

Final design - My Listings page

Final design - My Listings detail page

 

Next Steps

  • Adding in new and different types of data (reports, snowflake, GCP)

  • Thinking through ways to be able to better communicate to the user the relationships between parent and child data so that there are less pop ups.

  • Start to teach the user that there is a relationship and a hierarchy between certain data points

  • Incorporating a my access feature. User may not own some data but still want access to be able to use it.