Writing a post — the prompt and the voice.
Type a normal instruction and send. Your voice, your banned words and your content language are attached automatically — and Rewrite fixes the voice on any reply for free.
A post takes one sentence and about twenty seconds. Type what you want, hit send, read it back. Everything that makes it sound like you — your voice, your banned words, your language — is attached to the request before it ever reaches the AI. You don't configure any of it here.
The prompt
The box at the bottom of the thread. Write a normal instruction in normal English:
Write a LinkedIn post about our new same-day boiler
service. Mention the 2-hour callout window. Confident,
not salesy. End with a booking prompt.
- 4,000 characters is the limit. That's plenty for a full brief — an outline, a few facts, a customer quote you want worked in. Go over it and the send is rejected before anything is generated.
- The whole thread is one conversation, but each send is one generation. If the first draft is close, reply with what's wrong rather than rewriting your prompt from scratch.
- You can attach images — up to six per message. Attaching images →
- Cmd+Enter or Ctrl+Enter sends. Enter alone gives you a new line.
The brand chip
Bottom-left of the composer, showing the brand's logo and name. It tells you which brand the thread is writing as — and everything in the reply is derived from that brand.
Clicking it opens your brand list. Picking a different one on a thread that already has messages opens a confirmation, because a thread belongs to one brand permanently. Confirming switches your active brand across the whole app and starts you on a fresh thread. The full rule →
What gets applied automatically
Every message you send is wrapped in everything ScoutRival knows about the brand. You never paste any of this in yourself.
| What's applied | Where it comes from |
|---|---|
| Name, website, industry, About | Brand → Profile |
| Palette and fonts | Brand → Visual Identity |
| Tone, formality, sentence length, pacing, point of view, signature moves, do's and don'ts | Your trained voice fingerprint |
| Words we must never use, and words to favour | Your phrase rules |
| The language the post is written in | Your brand's content language |
If your output reads generic, it's almost always because your voice hasn't been trained. Paste three or four things you've actually written under Brand → Voice. Without a fingerprint, ScoutRival falls back on your industry — and industry-level writing is, by definition, writing any of your competitors could have published.
Two more things worth knowing:
- Banned words are hard rules. They're passed as “never use”. Preferred words are a preference, honoured when they fit naturally.
- Content language is set per brand, not per message. Set your brand's content language to Spanish and every Compose reply for that brand comes back in native Spanish — proper nouns, brand names and URLs excepted.
Rewrite in your voice
Under any reply, the Rewrite action re-runs that text through a dedicated brand-voice pass. It isn't a regeneration — it keeps the same ideas, in the same order, and changes how they're said.
It's shown a handful of your actual writing samples as examples, alongside your fingerprint and your phrase rules, and asked to match them. Showing a model two or three real examples is far stronger than telling it “be conversational”.
When it comes back, the reply is swapped in place and a voice-match score appears under it, with a line of advice:
| Score | Read it as |
|---|---|
| 85 and above | Sounds like you. Ship it. |
| 70–84 | Close. Worth a read-through before you publish. |
| Below 70 | Off. Usually a thin fingerprint — add more samples. |
- Rewrite is free. No credits, however many times you use it.
- It needs something to work with — a reply under about thirty characters is refused.
- The swap is local to your screen. The thread's saved history keeps the original message. Copy or Save the version you want before you navigate away.
There are no A/B/C variants
Some tools hand you three drafts and make you pick. ScoutRival doesn't. You get one reply, and two ways to change it:
- Send again — reply in the thread with what to change. This is a new generation, and it costs 2 credits like any message.
- Rewrite — keep the content, fix the voice. Free.
In practice, one good prompt and one reply beats three mediocre drafts you have to compare.
The other actions on a reply
- Copy — the whole post to your clipboard.
- Save — keeps the reply as a draft, titled from its first line.
- Make graphic — turns the post into a brand-locked image prompt and opens Prompt Studio with the headline already pulled out of it.
Getting better output
- Fill in Products & Services. Every idea has to map back to something you actually sell. Products & Services →
- Train the voice. Three or four real samples. It is the highest-leverage thing on this list.
- Give it a fact to hang on. A price, a guarantee, a callout window, a number of years. AI writing reads generic when it has nothing concrete to say.
- Say what you don't want. “No emoji”, “no rhetorical questions”, “don't open with a statistic” — these are respected.
- Use a preset when you write the same shape of post regularly, so you're not retyping the format instructions every week. Presets →