Hook prompts · First-line generation

10 AI prompts for tweet hooks that stop the scroll

The first line decides whether a tweet works. These 10 hook prompts produce specifically-engineered openers using the 5 structural patterns that consistently earn engagement — copy-paste for ChatGPT, Claude, Gemini.

A hook isn't a 'clever line' — it's a structure. The 5 patterns that work: curiosity gap, contrarian claim, specific shocking number, hidden-cost framing, personal stake. The prompts below generate hooks using each pattern, with variables to fill in your specifics.

The prompts

Prompt #1

Curiosity gap hook

Generate 5 tweet hooks (first lines only, max 130 chars each) using the curiosity gap pattern. Topic: {topic}. Pattern: state something that makes the reader want to know more without revealing the answer. Format examples: 'Here's why X is wrong about Y', 'The thing nobody tells you about Z', 'After {N} years of X, I figured out Y'. Avoid generic curiosity ('You won't believe this') — be specific.

Why it works

Curiosity gap is the most-used hook pattern. The 'avoid generic' constraint prevents the typical clickbait-AI output. 130-char limit forces sharpness in the opener.

Best for

GPT-4 (best at generating variants).

Variables to fill in

{topic}

Prompt #2

Specific shocking number hook

Generate 5 tweet hooks using the specific-number pattern. Topic: {topic}. Pattern: lead with a specific number (dollar amount, percentage, count) that's surprising or counterintuitive. Format examples: 'I spent $X on Y. Here's what I learned.' '{N}% of {audience} don't realize {fact}.' Max 130 chars each. The number must be specific (not '$1M+' or 'many') and credible (don't fabricate research statistics).

Why it works

Specific numbers signal credibility (readers don't fact-check; they assume specificity = real). The 'don't fabricate' guardrail prevents AI from inventing research numbers, which can hurt trust when readers do check.

Best for

Claude (best at the credibility filter).

Variables to fill in

{topic}

Prompt #3

Contrarian opener

Generate 5 tweet hooks using the contrarian pattern. Topic: {topic}. Pattern: state a widely-held view, then immediately push back. Format examples: 'Most people think X. Here's why they're wrong.' 'The standard advice for X is wrong for most people.' Max 130 chars each. The contrarian view should be one a thoughtful person could defend with evidence, not a hot-take-for-the-sake-of-it.

Why it works

Contrarian hooks trigger immediate reply velocity (people defending the consensus). The 'thoughtful person could defend' filter prevents flame-bait that hurts long-term account quality.

Best for

Claude 3.7+ (best at nuanced contrarian angles).

Variables to fill in

{topic}

Prompt #4

Hidden-cost opener

Generate 5 tweet hooks using the hidden-cost pattern. Topic: {topic}. Pattern: identify a hidden cost or opportunity people miss. Format examples: 'The hidden cost of X is Y.' 'You think X is the cost. It's actually Y.' 'People focus on X. The real expense is Y.' Max 130 chars each. The 'hidden' element should be specific and verifiable, not a vague concept.

Why it works

Hidden-cost framing forces the reader to confront something they overlooked. Specific + verifiable filter prevents 'the hidden cost is stress' generic AI output.

Best for

Claude (best at finding genuine hidden costs).

Variables to fill in

{topic}

Prompt #5

Personal stakes hook

Generate 5 tweet hooks using the personal-stakes pattern. Topic: {topic}. Pattern: lead with a specific personal experience that has high stakes. Format examples: 'I lost {amount/thing} doing X. Here's the lesson.' 'I rejected {opportunity}. Here's why.' 'I quit {job/path} at {age}. Best decision I made.' Max 130 chars each. Stakes must be specific (dollar amounts, age, named opportunities).

Why it works

Personal stakes earn empathy and curiosity. The 'specific stakes' filter ('$87k' beats 'a lot of money') makes the hook feel real rather than aspirational-influencer-fabricated.

Best for

Claude (best at the personal voice).

Variables to fill in

{topic}

Prompt #6

Insider revelation hook

Generate 5 tweet hooks using the insider pattern. Topic: {industry/role}. Pattern: open by establishing insider credibility, then promise a non-obvious insight. Format examples: 'After {N} years in {industry}, here's what nobody tells you.' '{Role}s know this; {target audience} don't.' '{Insider observation} is the thing that surprised me most.' Max 130 chars each.

Why it works

Insider hooks borrow authority. The credibility-establishment + 'non-obvious' filter forces the AI to think about what's NOT obvious (vs. what's general knowledge).

Best for

Claude or GPT-4. Both produce solid output with this structure.

Variables to fill in

{industry/role}

Prompt #7

Question hook

Generate 5 tweet hooks that open with a question designed to make the reader pause. Topic: {topic}. Pattern: the question should be open-ended, have multiple valid answers, and connect to a specific situation. Format examples: 'What's the biggest mistake you'd make as a beginner X?' 'How do you know when to quit X?' Max 130 chars each. Avoid yes/no questions — they don't generate continued reading.

Why it works

Question hooks reduce psychological resistance (readers can 'just see' what the answer is without committing). The 'avoid yes/no' guardrail is important — yes/no closes the reader rather than opening them.

Best for

GPT-4 (good at audience-relevant questions).

Variables to fill in

{topic}

Prompt #8

Counterintuitive observation hook

Generate 5 tweet hooks using the counterintuitive observation pattern. Topic: {topic}. Pattern: state something that runs against intuition but is true. Format examples: 'X looks like Y but is actually Z.' 'The opposite of common advice on X is what actually works.' Max 130 chars each. The observation should be one that would surprise someone with 1-3 years of {topic} experience.

Why it works

Counterintuitive observations force the reader to re-read. The 'surprises someone with 1-3 years experience' filter avoids both 'too obvious' and 'too niche'.

Best for

Claude 3.7+ (best at counterintuitive angles).

Variables to fill in

{topic}

Prompt #9

Loss/failure hook

Generate 5 tweet hooks using the loss-or-failure pattern. Topic: {topic}. Pattern: lead with a specific loss or failure you experienced and tease the lesson. Format examples: 'I failed at X. Here's what I learned.' 'My biggest mistake in {topic}: Y.' 'Lost {amount} to {failure}. The lesson nobody warned me about:' Max 130 chars each. Specifics required: the failure must be named.

Why it works

Loss/failure hooks earn 2-3× the engagement of success-led hooks. Reader bias: failures feel more honest and educational than wins. Forcing the failure to be named prevents vague 'I struggled with X' output.

Best for

Claude (best at the failure narrative tone).

Variables to fill in

{topic}

Prompt #10

Inversion hook

Generate 5 tweet hooks using the inversion pattern. Topic: {common advice in your field}. Pattern: take a piece of common advice and invert it, then signal you'll defend the inversion. Format examples: 'Stop doing X. Do Y instead.' 'Everyone says X. Try the opposite.' 'The advice that hurt my {goal}: X. What helped: Y.' Max 130 chars each. The inversion should be defensible, not contrarian-for-contrarian-sake.

Why it works

Inversions earn engagement because they challenge assumptions readers don't usually question. The 'defensible' filter is critical — undefendable inversions damage account credibility over time.

Best for

Claude (best at the inversion that's actually defensible).

Variables to fill in

{common advice in your field}

Common questions

Why is the hook so important on X?+

Because X's algorithm scores engagement velocity — likes/replies per minute in the first 30 minutes. If the hook doesn't stop the scroll, no engagement happens, no velocity, no algorithmic amplification. The first sentence of a tweet decides whether the rest exists. Optimizing the hook is the highest-leverage edit you can make.

How many hook variants should I generate?+

5-10 per topic. Generate 5+ variants, then pick the strongest 1-2. Most prompts above generate 5 at a time — this is the right batch size for picking. Generating 1 at a time produces lazy output (no comparison set); generating 20 at a time produces fatigue (you stop reading critically).

Should I test multiple hooks for the same content?+

Yes for important content (launches, milestone announcements). Generate 5 hook variants, pick 2, post variants 4-7 days apart to different content. Track which earns better engagement. Over 10-20 tests, you'll learn which hook patterns work for YOUR audience. Generic 'curiosity gaps work' isn't enough — your specific audience has preferences.

Can a great hook save bad content?+

For a single tweet, yes — a great hook earns the click, regardless of payoff quality. For a thread or long post, no — readers who clicked through to bad content remember and stop trusting your hooks over time. The pattern that compounds: great hooks + great payoff. Don't optimize hooks while neglecting content depth.

Skip the prompt engineering

AutoTweet's AI generates tweets in your voice using these structural patterns automatically. No prompt-tweaking required. 14 voice-matched tweets queued the moment you connect X.

Cancel anytime