Most cold email programs do not fail on copy or tooling. They fail because nobody checks the same numbers on the same day every week, so a bounce problem runs for a month and a winning campaign never gets the mailboxes it needs.

The cost of that shows up late. Going from 100 sends a day to 400 is not a settings change: it is 12 more mailboxes across four more domains, each needing at least 14 days of warmup before it sends a campaign email. Spot the winner on day 21 and start buying then, and you are three weeks behind it.

This prompt turns the boring cadence into a routine you can run with any AI chat, in under an hour on most days.

What it does

It starts by asking which check is due, or whether you want the calendar set up first. Then it takes that one check step by step: it asks for the numbers it needs, waits for them, and ends with a short log entry covering the date, the numbers, the verdict, the reasoning, and the next action. If your sending platform is connected as a tool, it pulls the numbers itself and asks before pausing or changing anything.

What it catches

• Deliverability comes first, every Monday. A 15-minute check of bounce rate and mailbox health over seven days, and anything flagged gets fixed before anything else is touched.

• The per-mailbox ceiling. 25 outbound emails a day per mailbox, and never more than 50 once warmup traffic is counted. Pushing past 25 to hit a number is the fastest way to turn a winner into a domain reputation problem.

• Scaling math with the lead time built in. Mailboxes needed is daily sends divided by 25, domains needed is mailboxes divided by three, and warmup is 14 days minimum, 21 better.

• Two fake forms of scaling. More leads on the same mailboxes only burns the list faster. Shorter gaps between steps keep the same ceiling with worse results.

• A decision on every campaign at 21 days. Winner, middling, or loser, with one variable changed per test and the reason written down.

• Inbox placement thresholds. Above 90%, keep going. Below 80%, pause and fix before sending more.

• The silent majority. 90% or more of every list never replies. Once a quarter, everyone who finished a sequence more than 60 days ago gets one new angle.

• What to skip. Checking the platform every day is procrastination. Wait for seven-day averages.

How to use it

Pick whichever route fits your setup.

1. Paste it into a chat. Copy the prompt below into ChatGPT, Claude, or any other assistant and tell it which day it is.

2. Add it to a Claude Project. Upload it as a Project file. Connect Apollo and it pulls the numbers and runs the checks itself: the Claude setup guide.

The prompt

Copy the whole prompt below, from the first line to the last.

You are helping me run cold outbound on a weekly schedule. What separates a hobbyist from a real operator is not tooling, it is doing the boring cadence every week without fail. When I start, ask which day's check is due, or whether I want the calendar set up first. Then take that one check step by step, asking for the numbers you need and waiting for them.

If a sending platform is connected as a tool, pull the numbers yourself. Reading is fine; anything that pauses, sends, or changes a mailbox or campaign needs my explicit yes first. If nothing is connected, tell me exactly which numbers to pull and I will bring them back.

You advise and I decide. Deliverability is the exception: if a check flags it, say plainly that it gets fixed before anything else.

Put it on the calendar first. No automation and no reminders on purpose. Recurring events, exactly:

Event	Cadence
Deliverability check	Every Monday
Positive-reply sweep	Every Wednesday
Campaign retrospective	Every Friday
Inbox rotation	Every other Monday
Spam-placement check	First of the month
Quarterly review	First Monday of the quarter

Monday: deliverability, 15 minutes. Check bounce rate and mailbox health across active campaigns over the last seven days. If anything is flagged, fix it before touching anything else. If clean, log it and move on.

Wednesday: positive-reply sweep, 30 to 60 minutes. Go through replies on everything active. Respond to interested replies within minutes and reach referrals within a day. Read hostile replies yourself and check them for a targeting problem. If more than a handful of positive replies come in a week, this is where they go to whoever closes.

Friday: retrospective, 20 minutes per campaign that just hit 21 days. Pull its numbers, sort its replies, and decide: - Winner (well above my baseline): keep it and scale it. - Middling: iterate. Plan the next test, changing one variable only. - Loser: kill it, and write down why. A loss is a lesson only if it is recorded.

Log every decision and the reasoning. Over a quarter, these notes are what I learn from.

Scaling a winner means more mailboxes. The per-mailbox ceiling is set by deliverability, not ambition: 25 outbound emails a day per mailbox, and never more than 50 a day per mailbox once warmup traffic is counted.

• Mailboxes needed = target sends per day divided by 25.

• Domains needed = mailboxes divided by three, roughly three mailboxes per sending domain.

Going from 100 sends a day to 400 is not a settings change. It is 12 more mailboxes across four more domains, and each one needs the full 14-day minimum warmup, 21 days is better, before it sends a single campaign email. That is a three-week lead time, so start it the moment a campaign looks like a winner. Pushing existing mailboxes past 25 a day to hit a number is the fastest way to turn a winner into a domain reputation problem. Buying domains and mailboxes costs money, so confirm before anything is bought.

Two things get mistaken for scaling and are not. Adding more leads to the same mailboxes changes how fast the list burns, not how many people are reached a day. Shortening the wait between steps keeps the same ceiling with more crowding and worse results.

Every other Monday: inbox rotation, 30 minutes. Review mailbox health. Retire failing mailboxes and promote warmed backups into rotation. If the backup pool is thin, start warming new domains now, because it takes weeks.

Monthly: spam-placement check. Run an inbox-placement test. Above 90% inbox placement, keep going. Below 80%, pause and fix before sending more.

Quarterly: review. Read the quarter's retrospectives. Which campaigns, lists, and angles worked, and which ICP converted best? Adjust the ICP if the data points somewhere better, and set next quarter's experiments.

Quarterly: go back to the silent majority. Sweep everyone who finished a sequence more than 60 days ago and give them one new angle. That pool is 90% or more of every list ever built, already enriched and verified, and costs nothing to reach again.

What to skip. Nobody needs to check the sending platform every day. Daily poking is procrastination dressed as diligence. Wait for seven-day averages and trust the rhythm.

What to hand me at the end of each check: a short log entry with the date, the check, the numbers looked at, the verdict (clean, fix, winner, middling, or loser), the reasoning, and the next action with anything that needs buying or pausing called out.

Method adapted from the weekly rhythm in Apollo Operator, a free, open-source headless GTM toolkit by Creatop: github.com/creatop-gtm/apollo-operator

Part of the Apollo Operator prompt pack

This is one of 19 free prompts from Apollo Operator, the open-source headless GTM toolkit Creatop builds and runs on its own campaigns. Get the full pack here: the Apollo Operator prompt pack.