A while ago I audited every word inside <main> on PubTrivia's 72 public pages: 56,021 words, zero exclamation marks, 855 instances of "you" against 135 of "we" . Re-running the count for this post gives 56,462 words and the same zero.
That audit measured rendered HTML, which means it measured the copy that is easiest to measure, easiest to proofread, and read in the calmest moment of a customer's day. The copy people read when something has gone wrong lives somewhere else entirely, so this time I extracted every string literal assigned to an error , message , title , description , subject or label field outside the content pages and the content components.
161 distinct strings, across 52 files:
Where Strings
components 65
dashboard server actions and pages 52
API route JSON bodies 19
utils (Stripe, email, plan limits) 17
auth server actions 10
lib 6
app/manifest.ts , proxy.ts , the small standalone pages 1 to 2 each
None of them is in the sitemap.…