Ideas on producing efficient, easy-to-read dashboards: how building rules around their creation is key to optimizing their performance. Working from ideation, to Proof-of-Concept and on to a well-used dashboard in Production.
Presented at the Dublin Tableau User Group meeting, by Mark Kernke, on 1st June 2016.
3. EXECUTIVE DASHBOARDS
PURPOSE
KPI Performance - Tracking to expectations for Quality and Quantity
Cross-Functional Reports - Where everyone is responsible
Readable at a Glance - Think Mobile
6. CHALLENGES
PROJECT MANAGEMENT
1. Start With a Sketch
Drawing your dashboard will help identify all teams and stakeholders for
Input and Approval
2. Map the Process
For every element, assign an owner for Input and Approval and keep a
checklist for state (Able to access data, Have targets, Have QA / Approval)
3. Parallel Processing
Don’t work on just one at a time, let people prepare for you and maintain
state centrally with last contact, ETA and any blockers or confusion
7. CHALLENGES
DATA DEFINITIONS
Transparency
Embed a “Plain English” Glossary of definitions and link to the current working
ETL processes (ideally in version control)
Ship Fast
Align with current Operational Reporting before aligning definitions across teams
8. CHALLENGES
DATA STRUCTURE/SOURCE
Fancy Counting - Deciding how you need to calculate a metric can change
(cumulative / incremental / ratio / rolling window)
ETLs will Iterate - Two people can use the same SQL and get two different
results…. somehow
Unify Datasources - Focus your data in one place so it is consistent
Solution: Manually dump into a Google Spreadsheet until final approval
12. DASHBOARD CONSTRUCTION
SECURITY
Data Source User Level Filtering - Which filters should be applied to data
sources on a per user basis (very manual)
Worksheet User Level Filter: More places to maintain, but has Active
Directory Group options (does stop users with Web Edits or Download
enabled)
Project Level Permissions: Those who have access, have access to everything
14. THE FUTURE AT GROUPON
STANDARDS AND PROCESSES
GRAAF - Groupon Reporting and Analytics Audit Framework
Review of architecture and choices after migration to Production Server
Alignment in Data Sources and Definitions with Data PM and Function Heads of
Analytics
Optimize database performance with full ETL review by DBA team
Stronger visualization style from Data Visualization
15. Recap
Key Takeaways
Know your stakeholders - For Input and Approval
Deliver before optimizing - Manual to Automated, Data Definitions, etc
Be Transparent and Collaborative - Documentation and Source Control
Simple designs and tools move faster - Keep It Simple Stupid