Results from testing thematic clustering vs. random SRS reviews

I recently switched from random SRS decks to thematic clustering—grouping terms like household tech or historical events. Before, I would hit a mental wall after 15 minutes. By testing this context-heavy approach for two weeks, my retention per session jumped by 30% without the usual burnout. Has anyone else experimented with sorting cards by scenario rather than frequency? Looking for more ways to keep reviews engag