Playbook replay
Brand voice compliance check
Five live captions scored against the voice doc, with every off-voice line named and the two worst rewritten.
Run for Lumira, a demo brand we invented for this page. No customer data. Nothing here runs against a live account.
Every session here was run once and saved, against a demo brand or public data. On your own account Naise works from your brand voice, your campaigns and your connected channels, so it reasons about your data instead of a stand-in.
A voice audit with no source document is just an opinion. The score only means something if every deduction points at a named rule.
Reading the voice doc first so the score is against your rules rather than my taste, then going line by line.
Overall: 6 out of 10.
That is a middling score, and the reason is consistent rather than random: the captions are on-voice when they talk about routine, and off-voice the moment they talk about results.
| Caption | Score | Flags |
|---|---|---|
| 1, morning routine | 9 | None |
| 2, ingredient explainer | 8 | One hedge: "may help to" |
| 3, results post | 3 | Banned term "flawless", clinical claim with no trial reference |
| 4, promo | 4 | Third person, 31-word opening sentence, CTA not in the library |
| 5, community repost | 7 | Signature phrase used, but sign-off is generic |
The pattern worth fixing. Every deduction sits in a results or promo caption. Those are also the two written under time pressure, which is the actual root cause: the voice holds when there is time to edit and collapses when there is not. A pre-approved results template would fix more than a rewrite will.
Rewrite: caption 3, the results post
Was: "Get flawless skin in 12 weeks. Clinically proven to reduce fine lines and even out tone."
Breaks two rules: "flawless" is on the banned list, and the second sentence is a clinical claim with no trial reference.
Now: "Twelve weeks. Same lighting, no retouching. Texture softened and redness calmed. Trial notes in our bio if you want the numbers."
Rewrite: caption 4, the promo
Was: "Lumira is pleased to announce that customers can now enjoy a limited-time offer of twenty percent off across the entire range for a limited period only."
Breaks three: third person, a 31-word opening sentence against a 14-word guide, and a CTA that is not in the library.
Now: "Twenty percent off, this week only. The whole range, including the refills. Shop the link in bio."
That is 18 words against 31, second person throughout, and the CTA is the library's standard one.
Naise created an artifact
Per-caption score with every deduction tied to a named rule, plus two rewrites.
Scores
| Caption | Score | Main flag |
|---|---|---|
| Morning routine | 9 | None |
| Ingredient explainer | 8 | Hedge: "may help to" |
| Results post | 3 | Banned term + unreferenced claim |
| Promo | 4 | Third person, 31-word opener |
| Community repost | 7 | Generic sign-off |
Root cause
Every deduction is in a results or promo caption, both written under time pressure. A pre-approved results template fixes more than a rewrite.
That was a sample. Run it on your own brand.
Naise reads your brand, your channels, and your market, then does this against your real data.
Run this playbook free