My Tableau report has 0.7M rows and 60+ fields, causing slow performance on Desktop and Server. I want to build a self‑service dashboard where users can add/remove fields as needed. Looking for guidance on steps to design this efficiently.

Your issue is due to poor data model + heavy live query design, not just dashboard design. You need to optimize the data first.

Correct Approach

1. Use Extracts (not live connection)

• Create a Hyper extract
• Reduce data at extract level (filters, aggregation)
• Enable extract optimization + aggregation

2. Reduce Data Volume

• Do not load all 60+ fields by default
• Keep only required fields in base dataset
• Create multiple smaller logical datasets if needed

3. Use Context + Data Source Filters

• Apply data source filters (date, region, etc.)
• Use context filters to limit data early

4. Build Self-Service Properly (Key Point)

You cannot let users truly “add/remove fields dynamically” like a BI tool UI, so:

• Use: 
  • Parameters + calculated fields
  • Show/Hide containers
• Or provide: 
  • Multiple prebuilt views instead of one heavy dashboard

5. Optimize Calculations

• Avoid: 
  • Complex row-level calcs
  • Too many LOD expressions
• Pre-calculate in ETL / Prep layer

6. Limit Marks & Visual Load

• Keep charts lightweight
• Avoid showing hundreds of thousands of marks

Recommended Architecture

• Tableau Prep / ETL → clean, aggregated dataset
• Tableau Extract → optimized
• Dashboard → modular views + parameters (not dynamic fields)

Final Conclusion

• The problem is data model + volume, not just UI
• True self-service = controlled flexibility (parameters, multiple views)
• Best solution: optimize data + use extracts + limit fields + modular design

You cannot make a single heavy dataset fully dynamic —