I build a one-page voice brief before translating copy and it saves me from faithful-but-flat results. Mine has 3 anchor verbs, 2 tone don’ts, and a target sentence length I sanity-check in Hemingway; CTAs get a quick back-translation to confirm they still sell. Got any templates or tools that help you guard brand promise on CTA-heavy pages?
Want a template? I use a CTA matrix per locale with approved verbs, banned tones, T/V level, char limits, and a pattern like ‘verb-first, benefit-second,’ wired into the style guide in Phrase so QA flags off-brand CTAs; the back-translation sits next to a one-line ‘reason to click,’ which works like a seatbelt for tone.
, button truncation drives me nuts, so I bake a layout check into the brief: I paste each locale’s CTAs into a mobile frame and run pseudo‑localization to catch overflow before QA. Hemingway’s fine for paragraphs, but for CTAs I do a 5‑second test on the back‑translation in Lyssna to see if people still read the intent instantly: https://lyssna.com/. If you’re using ‘3 anchor verbs,’ do you set a max character budget per verb per locale, or is that overkill?
I add a “click promise” column to the one‑page brief: for each CTA I jot the post‑click outcome (modal, checkout step, doc) and sanity‑check your back‑translation against that promise. It only adds a minute per CTA but catches those faithful‑but‑flat moments where, , tone sells the wrong thing; if the promise shifts, I swap the verb or add a tiny qualifier. Curious if you or @harper_j57 have a lightweight way to log destination types, or do you just rely on memory?