Substitution ciphers hide which letter is which, but they can't hide how often each letter appears. English text is roughly 13% E and 9% T; those proportions survive encryption untouched. Frequency analysis (counting cipher letters and matching the counts to known language statistics) has been breaking these ciphers since the 9th century, and it's the first real technique every cryptogram solver learns.
ETAOIN SHRDLU, and what to do with it
The most common English letters, in order: E T A O I N S H R D L U. In a decent-length cryptogram, the most frequent cipher letter is very likely E or T. That single guess, tested against word shapes (does it make THE work? does a doubled pair become EE or TT?), starts the chain reaction that unravels the whole puzzle. Frequency gives you the hypothesis; the words confirm it.
Cipher Munch can show the count under every cipher letter in the grid, or as a frequency table below the puzzle: your choice in Options → Display. No hand-tallying on scratch paper: the analysis layer is just there, so you can spend your attention on the deductions.
It changes by language, and the app knows
Spanish frequency order begins E A O S R N I: vowels are more common than in English, and T drops far down the list. If you carry English instincts into a Xenocrypt, your first guesses will be wrong in a consistent, fixable way. Practicing Spanish puzzles with frequencies visible recalibrates your instincts fast, and the built-in reference sheet keeps both languages' statistics one tap away.
Where frequency analysis shines (and where it doesn't)
It's strongest exactly where other techniques are weakest: Patristocrats, where no word boundaries exist and counting is most of what you have. It's least useful for Caesar ciphers: one confirmed letter reveals the entire shift, so counting is overkill. Cipher Munch reflects that honestly: the frequency table is hidden for Caesar puzzles, and you get a rotating decoder strip instead, because that's the tool that matches the cipher.
The endgame: turning it off
Competition solvers don't get an app-computed table: they count. But here's the twist: after a few hundred assisted solves, you barely need to. You'll recognize an E-shaped distribution at a glance, notice the suspiciously rare letters (hello, Q and Z candidates), and feel when a puzzle's statistics are lying to you (short quotes get weird). At that point, turn the frequency display off in Options and enjoy the strange new sensation of doing the analysis by instinct. The tool taught you; now the skill is yours.