For me, it's the advice that Anki sentence mining cards should be either audio only or text only.
At some point, I realized that only giving myself one or the other was having more negative effects than positive. The audio-only cards, for instance, meant I wasn't really associating the written form of a word with its sound, so I'd often encounter words when reading that felt unknown until I'd hear them and go "OH RIGHT." Also felt like it wasn't actually training my listening as well as I thought it would because often I wasn't really catching all the words or grammar, but I kinda knew anyway through audio cues and context. Text-only cards meant I wasn't hearing a native intonation and just had to imagine it, and I'd just often get frustrated staring at kanji trying to remember what it sounds like even though I knew the meaning.
When I switched to putting audio and text on the front, it immediately meant I could take on way more new cards per day without getting overwhelmed, pass them more often, get frustrated way less, and my retention of kanji, new vocab and grammar points went up much faster. Nice thing is that if I still want to test just listening or reading, it's as simple as closing my eyes to listen to the audio or muting my device to focus on the written form.
What about you guys? Anything you look back on now and go "why the hell did I listen to that?"
by Big_Description538