I’m genuinely asking here. This isn’t bait or a challenge, and I’m not really sure how to avoid that tone so apologies if seems antagonistic.
Basically, quick arcade-style games that can give them "points" so they feel good about just doing problems over and over for repetition. I can handle the conceptual discussions myself, but I find that gamifying works well for the portions where there's no alternative but to just practice until memorized.
I had also been working on a Treasure Mountain/Mathstorm/Galaxy clone with expanded activities, but I think the arcade format actually works better and is easier to make.
My daughter can also read a couple years ahead, but her writing's still very basic/phonetic. So I had codex whip up a standard template where I can put a picture (that codex generates) and she can write about what's happening in the picture and then color it. The template includes a word bank where I can add words relevant to what's happening (e.g. that I used to prompt the picture creation) that have irregular spelling, and I had it print a reminder sheet of different ways to spell the same sounds (ai, a_e, ay, etc) that she can refer to while trying to remember how to spell words. Previously when she was earlier in learning to read, I had used it to generate a couple simple coloring storybooks as a little fun activity.
That doesn't really answer your question about how I know it's effective, but I get a general sense that educational research struggles with that question too.