Yeah, WaniKani’s method is… kinda made up. A lot of the radicals that WaniKani uses were invented out of whole cloth as mnemonic components. Kanji dictionaries for Japanese speakers will sort kanji by their primary radical and then numbers of strokes, but don’t break the whole kanji down into components. For example, 銀 is the 金 radical plus six strokes, or 時 is the 日 plus six strokes, and so forth.
But! You do see the same components being repeated in kanji time after time. And because of that, it can be convenient to describe a kanji in terms of components. Breaking down 時 into 日＋寺 or even 日＋土＋寸 is a whole lot easier to memorise for learners than “oh yeah, it’s 日 plus a bunch of extra strokes”
So, why do we keep seeing the same components even though that’s not how kanji were formed? Sometimes it’s because they were, in a sense, assembled from components - a way you’ll frequently see that is with a semantic-phonetic pairing, where one part of the kanji (the semantic part) contributes the meaning, while the other part (the phonetic part) contributes the reading. About 80% of kanji are formed like this (though in modern Japanese, it can sometimes be a bit hard to tell, because the original formation of the kanji happened in the original Chinese, and so they occasionally no longer retain their original readings or meanings).
For example, the kanji 語 is comprised of 言 plus 吾 - the former is the semantic component and contributes the meaning “word”, while the latter is the phonetic component and contributes the reading ご. Interestingly, the example you use 外 was originally one of these as well - the semantic component wass 卜, representing an oracle bone, while the phonetic component was originally 月, which became 夕 over time.
The othe 20% of kanji are largely pictograms, which gradually changed from pictures into older forms of kanji called bone script, which got more and more refined until someone went “oh, these collections of strokes look kinda like kanji components that we aready have, so let’s just standardise them”.
Etymological drift, basically. Things change over time. Strokes move around because it’s easier to draw that way. For example the ⺌ radical you mention is actually a variant of 小. 氵 is the same radical as 水, 灬 is 火, ｲ is 人, 忄 is 心, the list is endless.
(And also sometimes it’s because WaniKani gives the same name to two etymologically different radicals, like 礻 and 衤 both coming under the heading of 礻 on WaniKani even though the former 示 is a variant of while the latter is a variant of 衣.)
that isn’t directly helpful for recognition/writing. – 
