Content classification
Providing Data Mapping capabilities for B2B customers through Smart Data governance & enhanced database visibility
#machinelearning
#database
#dashboards
Role
Research synthesis
Visual & Motion design
User testing
Product manager, Jesse
Product designer, Dimitar
Deliverables
Interactive Dashboards
Mini-design system
Duration
3 months
Customer problems & Project overview
Transcend customers house their customer data, on many internal databases like AWS, Redshift, Salesforce. The overarching problem we addressed was using data discovery, mapping and classification to centralize these data across their systems.
These scanned data silos and auto-classified personal datapoints enable customers to generate reports, populate and maintain an accurate, up-to-date data inventory.
Companies save time and resources–all of this is done through these automated scans, easy task delegation, & smart classification tech.
What did I learn?
Building flows which handles 3 main problems:
A way to browse the folder structure of the database to see what's in it.
A way to see all the personal data our system could find in the database in one view
A way to train the machine learning model on how to find personal data specific to the company's needs.
Discovering data collected by your Company. These come in as Data silos
The empty state's CTA guides the first user action.
See connected data silos
✅ Current actions running on the silos
✅ All data silo states
Filtering through Data silos
Filter by: ✅Data silo type ✅Percentage classified ✅Owner
Browse schema. The Data silo browsing experience
✅ Case 1 - User is opening a completely scanned Data silo
✅ Case 2 - User is opening a Data silo still scanning
✅ Case 3 - The scanned Data silo is empty
These cards show all the possible states of a Data silo
Datapoints tab
The training wizard
Now the scanning is complete, and all Datapoints are discovered, it's time to Train the machine learning model.
Transcend picks random columns, and wants you to confirm/deny the suggestions. How this works:
✅ Looks for patterns in the silo, and displays that match a data category
✅ Prompts you to confirm match or deny mismatch
Other states
✅ Case 1 - User attempts to Train a model while the Datapoints scan isn't complete
✅ Case 2 - User attempts to open a silo is discovered from Transcend's period scan
In the configuration tab, Silo discovery can be turned on so Transcend periodically searches
for new silos. These become available in the inactive tab
Final tab - Overview
This is the summary of the company data from all Data silos
Solution & impact overview
Currently 2/3 of all Transcend customers use this brand new feature, which was introduced just last year.