Step 2 of 3
Going pro (agency)
Train your Brand Voice in 30 minutes
The three fields that actually matter + 20 real-world examples of each, copy-pasteable.
What this guide covers
Jump to section
Why Brand Voice is the feature that compounds
Brand Voice is the difference between AI output you edit every time and AI output you ship. It is also what makes Clipflow sound different on every account — without it, we would be a template bank wearing different skins.
The trap: most users open Brand Voice, type "friendly and professional", hit save, and expect the magic. The AI cannot act on adjectives. It needs specific instructions — ideally with example phrasings. Spend 30 minutes on this once, save 30 minutes per post forever.
Boost your engagement with these actionable tips that will help you grow your audience! What are your thoughts? 💪🚀
Reads like every other tool
Nobody told me the algorithm doesn't care about effort — only about the first 3 seconds. Here's the fix that actually works:
First-person, contractions, specific
The four fields — what each one does
Settings → Brand Voice has four fields. Voice name is a label for internal reference (leave as "Default" unless you run multiple voices). The other three are the calibration signals the AI actually acts on.
Field 1 — Voice name
A label. Used when you run multiple brand voices (agencies do this per client). Solo creators: leave as "Default" and move on.
Field 2 — Tone & style
The most important field. Describe how you actually sound, with specifics. The in-app placeholder ("casual, witty, direct, educational, hype-free...") gives the format — take it as a starting list, then expand with 3-5 mechanical choices.
Tone & style — the right way
casual, direct, hype-free. I write in first person and use contractions. My sentences are short, often under 10 words. I use em-dashes instead of commas for emphasis. I never start sentences with "So," — it reads as filler. I end short-form posts with a specific action ("try this next week") not a question ("thoughts?").
Friendly but authoritative. Approachable. Confident. Authentic to my audience.
Field 3 — Words / patterns to avoid
The AI has default patterns — overused words and filler constructions it reaches for when unsure. This field kills those patterns for your account specifically. Be specific; generic bans do nothing. The in-app placeholder ("jargon, passive voice, emojis, overly salesy language") is a hint, not a full list — expand it.
Words / patterns to avoid — the right way
"leverage" as a verb, "unlock", "game-changer", "actionable insights", "synergy", "deep-dive", "crush it", any sentence starting with "In today’s fast-paced", engagement-bait questions ending in "thoughts?"
Corporate speak, buzzwords, clichés.
Field 4 — Example hook
One actual hook from your published content that landed. This is the calibration signal — the AI uses it to match rhythm, sentence length, hook style. Pick your best-performing real hook, not a fantasy one. The in-app placeholder ("I spent 90 days doing X — here’s what nobody tells you about Y...") shows the shape we’re looking for.
Example hook — the right way
I burned $47K on my first SaaS before I figured out what nobody tells you about pricing.
Something catchy that grabs attention.
20 tone examples, by creator type
Pick the closest match, edit to fit. These are starting points — personalize them before shipping.
Solo creators / Influencers
- First-person, contractions, audience-direct ("you")
- Curiosity-gap openers over stat-drops
- Short sentences — under 10 words where possible
- End with a specific action, not "what do you think?"
- Use em-dashes for pacing instead of commas
- Third-person narrator ("creators often...")
- Professional-formal tone — it reads as distant
- Engagement-bait questions
- Emoji-heavy CTAs
- Starting sentences with "So,"
B2B / SaaS founders
- Specific numbers (revenue, user counts, timelines)
- Company names — yours and others
- Story-driven: "I / we built X and learned Y"
- Concrete moments, not generalized wisdom
- Second-person on LinkedIn, first-person on shorts
- "Leverage" as a verb, "unlock", "synergy"
- Vague "entrepreneurs must..." framings
- Hustle-culture language
- Guru-voice — "12 hard truths about..."
- Industry buzzwords without specific anchoring
Coaches / Consultants
- Framework-first — named concepts over vague observations
- Contrarian angles: "the thing nobody tells you about X"
- Numbered lists for teaching
- End with a concrete action, not a question
- Authoritative, not performative
- Engagement questions ("what resonated most?")
- Testimonial-style bragging
- Filler intros ("as a coach for 10+ years...")
- Emoji-heavy copy
- Generic motivational language
Podcasters
- Quote the host with em-dash attribution
- Surface ONE insight per clip, not three
- Soft CTA back to the full episode ("full conversation at the link")
- Reference the guest by name + role
- Preserve actual podcast phrasing where strong
- Summary-style captions (reads as a content mill)
- Selling the subscribe button
- Overly dense context-setting
- Breaking the conversation frame
- Adding takeaways the host never said
Avoid-list: the 15 words AI over-uses by default
These show up in AI output until you explicitly ban them. Copy this list as a starting point, then add your personal triggers.
Common AI default patterns (ban these)
- "leverage" as a verb — almost always replaceable with "use"
- "unlock" — overused to the point of meaninglessness
- "game-changer" / "game-changing"
- "actionable insights" — means nothing
- "synergy" / "synergize"
- "deep-dive" as a verb
- "at the end of the day"
- "moving forward" / "going forward"
- "low-hanging fruit"
- "circle back"
- "ecosystem" (when the thing is not an ecosystem)
- "streamline" without a specific before/after metric
- "empower" — applied to nouns that cannot be empowered
- "disrupt" / "disruptive"
- "mindset" — when used to avoid specific advice
Testing that your Brand Voice actually works
After saving, run this test: go to an existing draft, click Regenerate. Read the output. Three signals it is working:
- You can tell it is yours within the first sentence (not "sounds like a tool")
- Zero instances of the words you put in the avoid-list
- The rhythm matches your example hook (similar sentence length / cadence)
When to update your Brand Voice
Your voice evolves. So should the settings. Update Brand Voice when one of these triggers:
A new hook pattern is outperforming
If your last 5 top posts opened differently than your example hook, swap it in. The Brand Voice calibration follows performance, not hope.
You catch a filler word slipping through
Every time you edit AI output to remove a specific word, add that word to the avoid-list. Five updates and the AI never types it again.
Your audience changed
Pivoted niches? New target ICP? Full Brand Voice rewrite, then bulk-regenerate every existing draft against the new voice.
Try this workflow on your next recording.
Free tier, no credit card. Your first draft lands in about two minutes.
Start free on Clipflow