Hey!

  Although I'm still in the first half of the book, it seems to me some people are actively trying to complicate matters. Yeah, seriously, when creating material for entry into SRS software the goal is not to be as cryptic as possible.

I'll assume you're using Anki or something similar.
You might find the following works well:  Document well your cards. Make it as braindead as possible. Put the story and context on the front side of cards, as well as on the back side -- make little scenes with two parts or something -- but I would say to stay clear of adding random bunches of keywords and hoping to magically remember them. This way you can increase dramatically the amount of material you add per day without having to worry about recall rate. The idea makes sense if you think about it: it's better to know well 80% of ten thousand cards than 80% of a hundred cards. Additionally, if a card just doesn't stick, just remove it. Feel bad about it ? add five more to make up for it.

Here's an example:
===============
---front---
表示

My display is Hyooodji !!
---back---
ひょうじ
===============
Sure, the freaking answer is given right after the word, but I never had to think twice when I came across this word after adding the card.. in retrospect, this one kind of sucks though.

for kanji, it's the same thing. add sentences at the bottom of the keyword to clarify stuff.
See AJATT for more info, that's where all this comes from.

Good luck!

With the disclaimer that I'm not anywhere near through RTK1 yet (I'm only in the low 700's now), I find that when I'm having this sort of issue it nearly always has to do with the story not relating to the keyword in a way that's memorable for me.

I think for me there's a tendency to look at the primitives in a given character and make the story fit the characters somehow, often to the detriment of their relationship to the keyword.

What's working for me best the further along I go is to use whatever comes to mind most strongly in relation to the keyword and make the primitives and story somehow relate to that, rather than the other way around.

Since Heisig recommends reviewing from keyword to kanji, I guess this makes sense that stories made this way would stick better.

A second reason for failing a character on review is keywords that are similar in meaning to me. An example of this would be "distinguish" and "distinction." I had a problem with these two until I modified my "distinction" story to include someone exclaiming "dis stincs!"