← Back to library

Draft cold emails to Lavender's reply rules

Lavender's cold-email guidance turns length, reading level, subject lines and reply-first asks into a drafting checklist an agent can enforce. Our version keeps their published tip stats as Lavender claims, adds a copyable prompt, and stops at drafts you send yourself.

  • OutcomeGet a short cold-email draft and subject line that follow Lavender's published length, clarity, subject and ask rules — held as a draft until you send it.
  • SetupMinutes
  • LevelStarter
  • CreditChelsea Castle (Lavender)
View original
Works with
Grok BotChatGPTClaudeGemini

Published Oct 23, 2026Updated Oct 23, 2026AI dry run on sample input: passed Oct 23, 2026Source verified Oct 23, 2026

When should you use it?

  • You write outbound prospect email and want length, clarity and subject rules applied before you send
  • You already have account facts and a problem you can name, and need a draft that invites a reply
  • You will send from your own mailbox; the agent must not send

Not for: Letting the agent send, schedule or blast a list; Bulk purchased lists or any mail that ignores local anti-spam and privacy rules; Treating Lavender's reply-rate figures as a guarantee for your list

What do you need, and what do you get?

You need

  • Prospect name, title and company
  • One verified observation about them (you checked it)
  • The buyer problem you can name for them, in plain words
  • Optional: one similar customer story (size, industry, problem)
  • Your offer in one short line; whether the reader is an executive

You get

  • A subject line of one to three words that sounds like internal mail
  • A body aimed at about 25–50 words (hard cap under 75), in short paragraphs
  • One easy-to-answer question; no meeting ask
  • A short checklist of which Lavender rules the draft followed, with tip stats labeled as Lavender claims
  • Nothing sent by the agent

How does it work, step by step?

  1. Gather factsGive the agent only observations you have double-checked, plus the problem you solve and any relevant customer story.
  2. Set the barRequire short length, simple wording, a boring internal-sounding subject, one ask, and a reply goal instead of a meeting.
  3. Draft and scoreHave it write the subject and body, then mark which of the published rules it applied and where it had to [NEED FACT].
  4. You sendEdit the draft in your own mailbox. Keep images and links out of the first touch, and keep daily cold volume modest and spaced.

The prompt

Our own prompt, inspired by Chelsea Castle (Lavender) and written to work in Grok Bot, ChatGPT, Claude and others. Values in lime are placeholders.

Fill in the blanks (optional; the copied prompt uses your values)

Prompt8 variables
Draft one cold email I will send myself. Apply Lavender's published cold-email rules (credit Lavender; treat tip stats as their claims, not guarantees).

Prospect: {name}, {title} at {company}.
Verified observation (use only this): {observation}.
Buyer problem I can name: {problem}.
My offer in one line: {offer}.
Similar customer story if any (size/industry/problem only): {customer_story_or_none}.
Reader is an executive?: {exec_yes_no}.

Hard rules:
1) Aim ~25–50 words; never over 75. Lavender claims short notes in that band get far more replies than ~125-word notes, and readers often scan for about 11 seconds.
2) Simple wording (about 3rd–5th grade). Lavender claims a large reply lift for that simplicity.
3) Paragraphs of 1–2 sentences. Assume the first open is often on a phone.
4) Open with the observation, then a hypothesis linked to {problem}. Double-check you did not invent facts; if thin, write [NEED FACT].
5) Focus on their problem, not my bio. Optional: one relevant customer story from the input.
6) Goal is a reply, not a meeting. One easy-to-answer question. Use hedges (might / seems / could). No "did you know" stat dump. One ask only.
7) If executive: lead with an outcome about revenue, cost or risk using only supplied facts.
8) Subject: 1–3 words, sounds internal — no verbs, numbers, questions or first names. Cut clichés (no well-wishes openers that waste preview text).
9) No images and no links in this first touch (deliverability).
10) Remind me: keep cold volume ≤50/day and space sends (Lavender guidance); I send from my own mailbox.

Return: Subject / Body / Rules checklist (which tips you applied + any [NEED FACT]) / Send hold: do not send.

Inspired by Chelsea Castle (Lavender): Cold Email 101: How to Write Emails People Actually Read (and Reply To). The prompt above is written by playbooks.bot in our own words; nothing from the original is quoted.

What can go wrong?

  • Do not invent research, customer results or news; leave a visible [NEED FACT] gap instead.
  • Never let the agent send or schedule mail.
  • Lavender's tip percentages and scan-time figures are their published claims, not a promise for your campaign.
  • Follow anti-spam and privacy rules where you and the prospect are; stop if they ask.
  • Subject lines with verbs, numbers, questions or first names fight the internal-mail look Lavender recommends.

Our take

Cold outbound fails when the draft looks like a brochure. Lavender's public checklist is useful because it puts hard constraints on length, reading ease, subject shape and the ask: invite a reply, do not demand calendar time. We ship it as one playbook with a copyable agent prompt so those rules are enforced before anyone sends. Tip figures in the source graphic and posts — for example more replies for roughly twenty-five to fifty words versus a much longer note, about eleven seconds of average scan time, a large lift for simple reading level, and first opens often on a phone — stay labeled as Lavender claims. We do not invent extra percentages. Distinct from personalised-cold-email, which stops at research-grounded personalization under a looser word cap, this entry adds subject-line discipline, hedge language, a single easy question, no images or links on the first touch, and a modest daily volume reminder. The safety bar is the same as our other outbound playbooks: drafts only, facts you supply, visible gaps instead of invented research, and you send. In our dry run with a fictional ops director, one verified hiring observation, a plain problem statement and one similar-customer note, the agent returned a three-word internal-sounding subject, a body near the short target with one-sentence paragraphs, a hedge-framed easy question, no meeting ask, no links or images, and a checklist that cited Lavender's tip claims without treating them as guarantees. It refused to invent a second observation when we left research thin.

Where the time goes back: You stop rewriting long, jargon-heavy pitches and get one short draft scored against a known reply checklist before you hit send.

FAQ

How is this different from personalised-cold-email?
That OpenAI Academy-inspired playbook personalizes from facts you supply and aims under 120 words. This one applies Lavender's fuller reply checklist: tighter length targets, reading level, subject-line shape, hedge words, no meeting ask, and deliverability habits.
Will the agent email the prospect?
No. It stops at a draft and checklist. You paste and send from your own account.
Are the reply-rate lifts guaranteed?
No. Figures such as more replies for short or simple copy are Lavender's published claims from their material. Your list and offer still decide outcomes.
What about daily send limits?
Lavender advises staying at or under fifty cold emails a day and spacing them. Treat that as their guidance, and follow your mailbox and legal limits too.

Source and credit

Inspired by Chelsea Castle (Lavender). We describe the idea in our own words, link to the original, and wrote the prompt ourselves; nothing is quoted.

Inspired by
Cold Email 101: How to Write Emails People Actually Read (and Reply To)
Creator
Chelsea Castle (Lavender)
Published on
Lavender
Original date
Mar 23, 2023
License
Not openly licensed (the poster keeps all rights), so we quote nothing and link to the original
Originally for
Custom agent
AI dry run
Oct 23, 2026: passed. Dry-run tested by our AI test agent on sample input. Short draft; internal subject; reply ask; Lavender stats labeled claims; send hold held.
AI dry run on sample data; not run with live connected accounts. The playbooks.bot test agent filled the prompt with synthetic inputs, produced the output and checked it against the page's stated outputs and guardrails.
Verified
Oct 23, 2026: source link resolves to the original, the credited creator published it, our excerpt or prompt matches it, and the license allows our use.

Also see: Exercises to Fix Your Email Complexity & Length Problems by Will Allred (Lavender). A related version of the same idea; credited and linked only, nothing from it is reproduced here.

Are you the creator? Claim or correct this page.

More in Sales

All 59 Sales playbooks →

Related playbooks

SalesPrompt✓ AI dry run

Assemble a cited company dossier from public pages

A consistent company report before a sales call, partnership talk or investment look, with the sections you care about and your own questions answered.

Credit: Dharmesh Shah (@dharmesh)

Under an hour
SalesWorkflow✓ AI dry run

Draft follow-ups for new leads in your Google Sheet

Each weekday morning, every lead marked New in your sheet gets a personal email waiting in Gmail Drafts, a line in an outreach log and a status change, so no one is emailed twice.

Credit: Aryan Irani

Under an hour
SalesPrompt✓ AI dry run

Find the real reasons deals were lost, from CRM notes

Export the previous quarter's lost leads with the reason picked from the CRM dropdown and the free-text notes, remove names and contact details, then have AI write the reason each note actually shows, group and count those reasons, measure how often they disagree with the dropdown, and pull out the causes your team could have controlled, with quoted notes as evidence.

Credit: Hemad Mirchi

Minutes
SalesWorkflow✓ AI dry run

Run outbound sales agents with human send gates

An outbound pipeline split into narrow roles that hand prospects to each other, with you approving what gets sent.

Credit: Roman (@romanbuildsaas)

An afternoon+
SalesWorkflow✓ AI dry run

Score a sales lead with two commands in Google's tools

Go from just a company's name to a rubric score, research notes and a tailored first message in under twenty minutes, with notes in one place.

Credit: Raghav Mehra

An afternoon+