When to use it
Use this when you're running outbound to a defined persona or segment and need a full cadence, not a single email, and you want each touch to earn attention with a fresh reason rather than guilt-tripping a 'just following up.' It earns its keep when you have enough volume in one segment to justify a reusable frame, and a clear single offer to organize the touches around.
Do NOT generate one sequence for 'everyone', a cadence built for VP of CS and SMB founders at once will be relevant to neither. Build one per persona. And do NOT ship the output as-is: the model produces a frame with personalization gaps you must fill per prospect; an un-personalized sequence sent at scale is just polite spam with five steps instead of one.
The principle: a sequence is a series of different reasons to care, spaced over time, not the same email sent five times with escalating desperation. The teams that win at cadence vary the ANGLE (pain, proof, a useful insight, a new stakeholder lens, a graceful break-up), escalate value rather than pressure, and treat the break-up email as a feature because it reliably surfaces people who meant to reply.
The prompt
Prompt, paste into Claude, ChatGPT, Gemini
You are an outbound expert who designs sequences that get replies without being annoying. Your rule: every touch must earn its place with a NEW angle. You never write a guilt-trip bump ('just following up', 'did you see my last email'), never use hype, and never invent results.
CONTEXT
Number of touches: {{NUMBER_OF_TOUCHES}}
Time window: {{TIME_WINDOW}}
Target persona / segment: {{PERSONA}}
What we do (value prop): {{VALUE_PROP}}
The core pain we solve: {{CORE_PAIN}}
Proof points available (fictitious-OK customers, results, content): {{PROOF_POINTS}}
Desired outcome of the sequence: {{DESIRED_OUTCOME}}
Channel mix (email-only, or email + LinkedIn + call): {{CHANNELS}}
TASK
Design a {{NUMBER_OF_TOUCHES}}-touch sequence over {{TIME_WINDOW}} where each touch uses a distinct angle and one ask.
METHOD
1. Assign each touch a DIFFERENT job before writing: e.g. (1) trigger/pain, (2) give-first useful insight or resource, (3) proof/social proof, (4) a different stakeholder or angle on the problem, (5) graceful break-up. Adapt the mix to the number of touches.
2. Space the touches sensibly across the window, front-load, then widen the gaps.
3. Write each email under 100 words, one idea, one low-friction ask.
4. Mark exactly where the rep must add a prospect-specific line with [PERSONALIZE: what to add here].
5. If CHANNELS includes more than email, insert a LinkedIn or call step where it makes sense and label it.
OUTPUT FORMAT (per touch)
Touch #, Day X, Channel, Job of this touch
Subject:
Body:
[PERSONALIZE: ...]
Then end with:
TEST PLAN: if reply rate is low, what to change first and why (subject lines, touch 1 angle, cadence spacing), 3 concrete experiments in priority order.
CONSTRAINTS
- Each email under 100 words. One idea, one ask.
- No two touches may share the same angle or reuse the same hook. If you find yourself repeating, change the reason to care.
- Banned phrases: 'just following up', 'just checking in', 'circling back', 'bumping this', plus hype words (unlock, leverage, supercharge, seamless, game-changer, revolutionize).
- Use only the proof I gave you. Never fabricate a customer or number.
- Plain, human language at a 7th-8th grade level.
- If CORE_PAIN or PROOF_POINTS is thin, tell me which one and how it weakens the sequence before writing.Run it from the terminal
$# confirm the persona and touch count are set in the prompt
$grep -E 'NUMBER_OF_TOUCHES|PERSONA' prompts/sequence.md
Number of touches: 5
Target persona / segment: VP of Customer Success at growth-stage B2B SaaS
$# build a 5-touch cadence for one persona via the llm CLI
$cat prompts/sequence.md | llm -m claude-sonnet-4-6 | tee out/sequence.txt
Touch 1, Day 1, Email, Job: open on the pain
Subject: scaling onboarding
Body: [PERSONALIZE: reference their funding/hiring] Most CS leaders hit a wall when
onboarding can't keep pace with new logos. We help fix that without adding headcount.
Touch 2, Day 3, LinkedIn connect, Job: low-pressure presence
Note: Enjoyed [their post], working with a few CS leaders on scaling onboarding.
Touch 3, Day 6, Email, Job: give-first
Subject: first-90-day benchmarks
Body: Put together a short benchmark on first-90-day onboarding. Want it?
Touch 4, Day 10, Email, Job: proof + stakeholder angle
Subject: 30% faster, no new hires
Body: A Series C team cut time-to-value ~30% without adding CS headcount.
Touch 5, Day 14, Email, Job: graceful break-up
Subject: should I close this out?
TEST PLAN
1. If opens low: rewrite Touch 1 subject.
2. If replies low: move the benchmark to Touch 2.
3. If meetings low: soften Touch 4 ask to 'a 10-minute look'.
$# five distinct angles, no repeated hooks; fill PERSONALIZE before sending
$
Variables to swap
| Variable | What to put there |
|---|---|
{{NUMBER_OF_TOUCHES}} | How many touches in the sequence (e.g. 4-6) |
{{TIME_WINDOW}} | Over what period to spread them (e.g. 14 days) |
{{PERSONA}} | The single persona or segment you're targeting |
{{VALUE_PROP}} | What you do and the outcome |
{{CORE_PAIN}} | The main problem this audience feels |
{{PROOF_POINTS}} | Results, customers, or content you can reference (fictitious-OK, but real to you) |
{{DESIRED_OUTCOME}} | What a successful sequence produces (e.g. a booked demo) |
{{CHANNELS}} | Email-only, or email + LinkedIn + call |
Worked example
Input
Filled-in example
NUMBER_OF_TOUCHES: 5
TIME_WINDOW: 14 days
PERSONA: VP of Customer Success at growth-stage B2B SaaS
VALUE_PROP: Onboarding platform that speeds time-to-first-value
CORE_PAIN: Onboarding doesn't scale as the team grows; first-90-day churn risk
PROOF_POINTS: A Series C SaaS cut time-to-value by 30%; a short first-90-day onboarding benchmark report
DESIRED_OUTCOME: Book a 30-min demo
CHANNELS: email + LinkedInOutput
What the model returns
Touch 1, Day 1, Email, Job: open on the pain
Subject: scaling onboarding
Body: [PERSONALIZE: reference their recent funding, hiring, or a new tier] Most CS leaders I talk to hit a wall when onboarding can't keep pace with new logos, the team scales but the process doesn't. We help fix that without adding headcount. Worth a quick look?
[PERSONALIZE: one line tying their specific growth signal to onboarding strain]
Touch 2, Day 3, LinkedIn connect, Job: low-pressure presence
Note: Enjoyed [their post / their take on X], working with a few CS leaders on scaling onboarding and thought it'd be useful to connect. No pitch.
[PERSONALIZE: the post or detail you're referencing]
Touch 3, Day 6, Email, Job: give-first
Subject: first-90-day benchmarks
Body: Put together a short benchmark on first-90-day onboarding across growth-stage SaaS, useful even if we never talk. Want me to send it over?
Touch 4, Day 10, Email, Job: proof + stakeholder angle
Subject: 30% faster, no new hires
Body: A Series C team standardized onboarding and cut time-to-value ~30% without adding CS headcount, which also calmed their CFO down on tooling spend. Happy to walk through how in 30 minutes.
[PERSONALIZE: name their likely internal blocker, CFO, RevOps]
Touch 5, Day 14, Email, Job: graceful break-up
Subject: should I close this out?
Body: No worries if the timing's off, want me to circle back next quarter, or is onboarding just not a priority right now?
TEST PLAN
1. If opens are low: rewrite Touch 1 subject, test 'onboarding that scales' vs a question format. Subjects gate everything downstream.
2. If opens are fine but replies are low: move the give-first benchmark to Touch 2, leading with value before any ask often lifts reply rate.
3. If replies come but no meetings: soften the Touch 4 ask from '30 minutes' to 'a 10-minute look' to lower friction at the conversion point.Tips to get more out of it
- The model defaults to repetition, explicitly demand a new REASON to care in each touch, not just new wording. If two touches feel similar, tell it 'touch 3 and 4 are the same angle; make one of them a genuinely different lens.'
- Fill every [PERSONALIZE] tag before sending. The sequence is a frame; the personalization is what makes it land. Sending the frame raw at scale is the fastest way to burn a segment.
- Keep the break-up email, it consistently pulls replies from people who meant to respond, and the 'is this just not a priority?' framing gives them an easy, honest out that often turns into a real answer.
- Match cadence to deal size: enterprise can stretch to 3-4 weeks with wider gaps; SMB needs tighter spacing or the deal goes cold. Tell the model the segment so spacing fits.
- A/B test subject lines first and in isolation, they drive opens, and nothing in the body matters until the email is opened. The built-in TEST PLAN orders experiments for you; run them one at a time.
- For a give-first touch, the resource must be genuinely useful and ungated, or it reads as bait. If you don't have a real asset, drop that touch rather than fake one.
- Reuse the strongest touch as the basis for a single cold email (use the cold-email prompt) when you want a one-off, and reuse a sharp line as an ad hook later.