Laravel performance tuning bottleneck?

We run 'Laravel Quick Fix & Consultation', a SaaS application that's seen significant user growth recently. While scaling, we've hit a critical Laravel performance bottleneck that's severely impacting user experience, particularly for our power users.

Specifically, our main dashboard's data aggregation queries are taking upwards of 7-10 seconds for users with extensive historical data. This involves complex Eloquent relationships across several pivot tables and polymorphic relations, making the data retrieval incredibly slow.

Here's what we've already tried:

• Aggressive Caching: Implemented robust Redis caching for frequently accessed aggregated data, with proper invalidation strategies.
• Eager Loading: Ensured all critical Eloquent relationships are eager loaded using with() and load() to mitigate N+1 query issues.
• Database Indexing: Added comprehensive database indexes to all foreign keys and frequently queried columns across all relevant tables.
• Raw SQL Optimization: For specific, highly complex reports, we've bypassed Eloquent and used DB::select() with optimized raw SQL queries.
• Profiling Tools: Utilized Laravel Debugbar and Blackfire.io extensively for profiling. While these tools consistently point to the database layer as the bottleneck, specific query optimization efforts haven't yielded the expected breakthroughs.

The core issue seems to stem from a particular polymorphic many-to-many relationship that, even with eager loading and proper indexing, results in a massive number of joins and intermediate table scans when aggregating across tens of thousands of related records. We're seeing GROUP BY operations on very large result sets becoming the primary bottleneck, even after pre-filtering as much as possible.

We're looking for advanced strategies beyond conventional caching and indexing. Has anyone successfully optimized extremely complex, deep polymorphic Eloquent relationships for large-scale data aggregation without resorting to complete denormalization or a full rewrite to a NoSQL solution? Are there specific Laravel techniques or packages (e.g., for materialized views, advanced query builders, or alternative ORM approaches for reporting) that could tackle this specific Laravel performance challenge more effectively?

Anyone faced this before?