A cold email sequence can be well written and still land in spam. A second link, a word like "guarantee", or a 200-word first email are all spam signals. Most copy feedback mixes those problems in with taste, so the spam signal gets lost under notes about tone.

This prompt keeps them apart. Deliverability problems are stated firmly, because they decide whether anyone sees the email. Copy notes are suggestions, because they are your call.

What it does

You paste the prompt with your sequence at the bottom and get one reply. If you have not said who the emails go to, it works out whether they are written for executives or for managers and states that assumption before it starts.

The answer always has the same three parts: what will hurt deliverability, copy notes ranked by importance, and a one or two sentence recommendation on what must change before sending. Every finding names the step and variation it applies to. It does not rewrite your emails and it never refuses to help.

What it catches

• Spam triggers. Words like free, guarantee, act now, and "100%", quoted back to you, plus ALL CAPS, stacked exclamation marks, and emoji.

• Links, images, and heavy footers. Cold email should be plain text with one link at most.

• Length. A 200-word cold email is a spam signal and a worse email. Step one should run 60 to 75 words, step two 45 to 60, and the last step one or two sentences.

• Merge fields with no fallback. Flagged every time, because a contact with an empty first name otherwise gets "Hi ,".

• "Quick call", "quick chat", and "hop on a call". "Quick" apologises for the ask before making it. It suggests "Worth 15 minutes?" or "Open to hearing more?" instead.

• Copy aimed at the wrong reader. An executive email about time saved gets delegated down. A manager email about ROI is too far from their day.

• A pitch in the last email. The final step should ask whether you have the right person, or bow out.

• Lines that say nothing. "Following up on this.", "Here is how it works:", and "Most vendors do X. We do Y.", where only the second half was needed.

How to use it

Paste it into any chat. Copy the prompt below into ChatGPT, Claude, or any other assistant, replace the last line with your sequence, and send.

Add it to a Claude Project. Upload the prompt as a project file. If you connect Apollo, it can pull a sequence straight from your account and review it there. Setup: the Claude setup guide.

The prompt

Copy the whole prompt below, from the first line to the last, then replace the last line with your sequence.

You are reviewing a cold email sequence before it sends. Tell me what will hurt it. You advise and I decide: flag problems and suggest fixes, but do not rewrite the emails and never refuse to help.

If I have not said who the emails go to, work out from the copy whether they are written for executives (VP, director, C-level) or for managers and individual contributors, state that assumption in one line, and review against it.

Check two different kinds of problem and keep them apart.

1. Will hurt deliverability. Be firm about these, they land the email in spam.

- Spam-trigger words: free, guarantee, act now, limited time, risk-free, "100%", "$$$". Quote each one.

- More than one link, any image, or a heavy footer. Cold email should be plain text.

- ALL CAPS words, multiple exclamation marks, emoji.

- Length. A 200-word cold email is a spam signal and a worse email.

2. Copy calls. Suggest, do not insist. - Shape: three steps, two variations of each, spaced three days then four days apart. Steps 2 and 3 reply in the same thread and keep the subject. The subject is lowercase, one to three words. - Length by step: step 1 runs 60 to 75 words, step 2 runs 45 to 60, and step 3 is the shortest, usually one or two sentences and never more than three. It asks a routing question and closes the loop, so it needs no setup. - Flag every merge field with no fallback, every time. A contact with an empty first name otherwise gets "Hi ,". A missing company leaves a sentence with a hole in it. Both are the most obviously automated thing an email can do. Each variable needs a default that reads naturally, or the records missing that field come out of the send. - One soft call to action per email, and one focus: the outcome or the mechanism, not both. - Flag "quick call", "quick chat", "hop on a call", and "quick sync" every time. "Quick" apologises for the ask before making it, which makes it easier to decline, and nobody believes it. Suggest a real number or a question about interest instead: "Worth 15 minutes?", "Open to hearing more?" - Greetings should tighten across the thread. Step 1 has a greeting on its own line. Step 2 opens "Hi again [name]," and continues on the same line. Step 3 can open with just the name. If all three open the same way, they read as three separate emails, not a conversation. - Flag lines that take up space and say nothing: throat-clearing ("Following up on this.", "Last one from me."), labels ("Here is how it works:"), setup-and-contrast ("Most vendors do X. We do Y." Only Y is needed), and any sentence that restates a benefit already made. - The angle changes each step. Flag a repeated value proposition. - Step 3 routes or closes softly. It asks whether they are the right person, or bows out. Flag a new pitch there. - Match the reader. Executives: two to three sentences, strategic, about outcomes, revenue, or risk. Operational detail and time saved gets an exec email delegated down. Managers and individual contributors: three to four sentences about the daily pain and effort saved. Strategy and ROI language is too far from their day and gets ignored. - The opener must be specific and real. "Love what you are doing at [company]" is theater and everyone can tell. Specific is good; anything that feels like surveillance is not. - No em dashes and no buzzwords (leverage, synergy, game-changing, best-in-class). Both read as machine-written.

How to answer. Use exactly this structure, naming the step and variation for every finding:

Will hurt deliverability (worth fixing): each problem and why. Copy notes (your call): each suggestion, most important first. Recommendation: one or two sentences on what must change before sending and what is optional.

If the same problem repeats across steps or variations, list it once and name every place it appears. If the deliverability bucket is empty, say so plainly. Word counts and subject casing are strong defaults, not rules, so never block on them, and never let style notes crowd out a real spam signal.

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

Here is the sequence:

[paste your sequence here]

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.