Does this caption actually sound like you?
Paste two or three captions you have written before. Paste the new draft you are not sure about. Get a 0–100 voice match score, the specific stylistic moves that drifted, and a rewrite that scores higher.
This is the public version of the drift score Marqeting runs on every draft generated inside the product.
The scorer compares the new draft against your samples on five axes: vocabulary overlap, sentence rhythm, opener patterns, emotional register, and the cliches the samples avoid. A draft that picks up new AI tells the samples never had — em-dash filler, “excited to announce,” chirpy emoji openers — loses points fast.
Verdict bands:
- 75–100: in-voice. Ship it.
- 50–74: borderline. Rewrite the worst sentence and reship.
- 0–49: drifted. Do not publish without rewriting.
Marqeting trains a full voice fingerprint from your post history (not three samples) and runs the score on every caption, carousel, blog, and reel it generates — before you ever see the draft. Drifted output never reaches the calendar.
Train your voice fingerprint free →