Find AI Prompts
Home›Sales›Cold Outreach Sequences
SalesProspecting & Outreach

AI Prompts for Cold Outreach Sequences

A cold outreach sequence is a small campaign: several touches, across channels, over two to three weeks, each with one job. The first email earns a reply or a read; later touches add a new angle, proof, or a graceful exit. Sequences fail when every touch repeats the same pitch, when they run too long, or when the opener could have been sent to anyone.

This page keeps the original four-touch prompt as the fast version and adds the architecture and review prompts around it: how to design the touches, channels and spacing for your segment, and how to check a draft before it goes out. Angles come from the research pages; this is where they become a sequence.

Before you use these

Have these ready to replace the highlighted [variables]:

The prompts

1. Write a four-touch cold email sequence

Best forA tight, ready-to-send sequence when you have the trigger and the proof point in hand.
Inputs needed
  • Job title and company type
  • Product
  • Trigger event
  • Proof metric
How to use itThe original prompt from this page. Fill the variables with a real trigger and a real metric; the constraints in the prompt do most of the work.
Expected outputFour emails with spacing, each under the word limit, one ask each, no filler.
Act as an SDR coach. Write a 4-touch cold email sequence to [job title] at [company type] selling [product]. Email 1: problem-led opener referencing [trigger event], under 90 words. Email 2 (+3 days): case study angle with one metric. Email 3 (+4 days): different pain angle, question close. Email 4 (+5 days): polite breakup. No 'I hope this finds you well', no fake flattery, one clear ask per email.

2. Design the sequence architecture

Best forDeciding touches, channels, spacing and the job of each step before writing any copy.
Inputs needed
  • Persona and segment
  • Angles available
  • Channel performance and constraints
  • Tier treatment level
How to use itRun before writing when you are building a sequence for a segment rather than a single prospect. The model produces the skeleton; copy follows it.
Expected outputTouch-by-touch architecture with channel, day, job, angle, ask, and exit rules, plus the reasoning and the variants by tier.
Act as an outbound program designer building the architecture of a cold sequence for [persona] at [segment].

Angles available: [ranked, with the fact each rests on]
Channels and constraints: [email, phone, LinkedIn, video, other — with reply rates if known, sending limits, compliance rules]
Tier: [A/B/C and the personalization depth allowed]
Buyer behavior: [when they are reachable, what they respond to]

1. Define the sequence: number of touches, total duration, and the exit rule (positive reply, negative reply, no reply after n touches). Justify duration against buyer attention and deliverability.
2. For each touch: day, channel, the single job of the touch (earn a read, add proof, change angle, invite a small step, exit), the angle used, the ask, and the length limit. Ensure consecutive touches do not repeat the same angle or ask.
3. Channel logic: where the phone enters, whether LinkedIn precedes or follows email, how a voicemail references the email.
4. Personalization plan: which touches carry researched lines and which are templated, given the tier.
5. Branching: what changes if the prospect opens repeatedly but does not reply, engages on LinkedIn, or forwards internally.
6. Measurement per touch: reply rate and positive reply rate by step, so the weak touch can be found.

Present as a touch table plus notes. Do not exceed [n] touches without a stated reason; more touches rarely fix a weak angle.

3. Review a draft sequence before sending

Best forCatching the habits that kill reply rates — generic openers, feature lists, multiple asks, length.
Inputs needed
  • Draft sequence
  • Persona
  • Angle and proof
How to use itPaste the draft. The model checks each email against specific rules and rewrites the weakest lines rather than the whole thing.
Expected outputEmail-by-email review with rule violations, specificity and clarity scores, rewritten openers and asks, and the send/hold verdict.
Act as a senior SDR manager reviewing a cold sequence before it goes live.

Draft sequence: [paste all touches with spacing]
Persona: [role, priorities]
Angle and proof used: [summary]

Check each touch against these rules and report violations with the line quoted:
- Opener references something specific and true about them, not about us, not flattery, not 'I noticed your profile'.
- One idea per email; one ask per email; the ask is small (a question, a yes/no, a 15-minute call).
- Length under [limit] words; subject line under [n] words and not clickbait.
- No feature lists; proof is one number with a source; no unverifiable claims ('leading', 'best').
- Consecutive touches change angle, not just wording.
- No guilt-trip follow-ups ('I've reached out several times…'); the breakup is gracious.
- Reads naturally aloud; no jargon the persona would not use.
- Compliance: opt-out language where required, no misleading subject lines.

Then: a specificity score (1–5) and clarity score (1–5) per touch; rewritten openers and asks for any touch scoring 3 or below; the two changes with the largest expected effect; and a send / revise verdict for the sequence.

4. Write a single researched first email for an A-tier account

Best forOne email, fully personalized, when the account is worth an hour of research rather than a sequence template.
Inputs needed
  • Account brief
  • The specific person and something they said or did
  • Angle and proof
How to use itUse for the accounts at the top of the tier list only. Give the model the research; it writes one email and shows which line carries the research so you can judge whether it is specific enough.
Expected outputOne email under 110 words with the research line marked, a subject line, the ask, and a note on what would make it stronger.
Act as a senior AE writing a single, fully researched cold email to [name, role] at [company].

Research: [account brief highlights; something this person said, wrote or did, with source and date; the trigger; their likely priority]
Angle: [the reason this matters to them now] Proof: [one number, one customer]
Ask: [the small next step]

Write one email:
1. Subject line: specific to them, under 6 words, no clickbait.
2. Opening line that references the research in a way only someone who read it could write — no flattery, no 'I noticed'.
3. The bridge from their situation to the problem we solve, in one or two sentences, in their language.
4. The proof, one sentence, with the number.
5. The ask: one question, easy to answer.
Under 110 words. Then: mark the line that carries the research; say whether it would still make sense if sent to a different company (if yes, rewrite it); and name the one additional fact that would make this email materially stronger.

Before you hit send

The review prompt catches these; this is the short version for the moment before a sequence goes live.

Related prompts

Logical next step

After this, most sales teams move on to Outreach Angles and Message Hypotheses.

B2B Sales Prospecting System ($39) — goes further: a repeatable outbound engine: research → sequence → objections → follow-up. See what's inside →
Get the free B2B Sales AI Starter Kit → Nine of these prompts as a target → engage → qualify workflow with an intake worksheet, delivered by email. See what's inside

All Sales prompts · Search the full library