Testing radical-grouping to fix phonetic word confusion

I've been experimenting with grouping kanji by radical to stop mixing up similar terms. I used to confuse words like 体温 and 院長 during fast recall sessions. After three days of radical-mapping, my error rate on these specific cards dropped from 40% to near zero. Has anyone else tracked their data on this? I'm curious if this outperforms standard mnemonics for tricky words like 勿論.