Cold email course: the psychology behind every email that earns money

The twelfth course, closing MONA's free AI course series: writing cold outreach emails, following a 4-hour curriculum that treats this craft as a behavioral psychology science rather than a writing talent. Why does an AI course series end with a course on… writing letters? Because it is the last piece of the picture: every machine you have learned to build (scanning prospects, personalizing, sending automatically) only amplifies the quality of the letter inside: a bad letter multiplied by a thousand is still a bad letter times a thousand. And one more reason, very much to our taste: this curriculum takes a contrarian view of using AI to write sales letters, and says up front that it recommends it very sparingly: read Part 7 and you will see why people who live off automation advise this, and why we nod along. In Vietnam, "cold email" also takes the form of Zalo, LinkedIn and Facebook messages: all the principles in this article apply to those channels.

Part 1: Seven psychological levers that get strangers to reply

The foundation of the whole course is seven classic principles of influence (in the Cialdini tradition), retold in a salesperson's language:

1. Reciprocity: like a restaurant offering candy before the bill. In a letter: give a useful finding ("your sales page is misconfigured here, it is probably leaking this much each month, let me show you how to fix it") without asking for anything.
2. Commitment, in small steps: nobody agrees to spend tens of millions on a stranger, but anyone can say yes to a 5-minute request. Ask for the small yes first, then raise it gradually.
3. Social proof: people follow the crowd. The working rule: specific numbers ("last week we did this for a company in the same industry, 15 minutes' drive from you") beat vague claims, and the closer to their industry and size the proof is, the more weight it carries with the recipient.
4. Authority: certifications, achievements, and even a confident writing voice with no hedging. An expensive detail: authority must match what the customer cares about: an impressive degree loses to a partner badge for the exact platform the customer is running ads on.
5. Similarity: find common ground, and better still, mirror the writing style of the recipient: if they write short, you write short; if they are casual, you loosen up.
6. Scarcity: limited spots, a limited deadline, but it must be archive footage (your calendar is truly full, the offer truly expires): customers' bluff detectors are more sensitive than you think.
7. Shared identity: same trade, same values, same kind of struggle: a deeper layer of similarity.

Part 2: The three pillars of a campaign: a one-sentence goal, the one-to-one frame, and a loop with numbers

• A goal in one sentence. Each email picks exactly one target: get a reply, get something viewed, or get a meeting booked. If you cannot sum up the goal in one sentence, you are not ready to write a campaign.
• The one-to-one frame. A sales letter today has to read like a real person messaging a real person. A quick test: hand the letter to an acquaintance. Do they think it is personal or mass-sent? Clear out the "company" signals: stock greetings, grand signatures, "we" instead of "I". A counterintuitive tip for the AI era that really works: deliberately leave a trace of a real person (a clipped sentence, a hurried typing style), because spotless polish now smells like a machine.
• A loop with statistical meaning. Treat each campaign as an experiment: each variant needs 500-1,000 emails to reach a conclusion, and judging after sending a hundred is guessing. The standard rhythm: run a few variants in parallel, set a fixed weekly session to read the numbers, kill the weak variants, multiply the strong ones; test very different versions at first, then fine-tune in smaller steps. One campaign in the curriculum climbed from a 3.5% reply rate to 10% with this loop alone, and another that its own author was sure would fail hit 15%: data over intuition, always.

Part 3: The four-step formula for a letter

The template for every letter in the course, four steps in the exact order the recipient reads them:

1. Personalization (1 sentence, 2 at most): the best return of the whole letter, because it is the only part certain to be read. The core technique is called "cold reading": a sentence that sounds like it knows the recipient well but is actually true for the whole group ("from the way your channel has been since the early days, you like to speak straight without beating around the bush": true of most decent content creators). Banned: empty machine-style compliments ("very impressed by your company's passion for process optimization": one look and you know it is a bot).
2. Who I am and why listen: answer the unspoken question "is this person a scammer?" One or two sentences of proof with numbers: "I am working with a company in your industry, and last month the system brought in this much."
3. An offer in the "X in Y, or else Z" pattern: state an observation that hits where it hurts, then make an offer that sounds "too good to be true" but comes with a safety net: "I commit to delivering X in 60 days, and if not, I keep working for free until it is done." The math behind it makes a big promise feasible: a number that sounds huge to you is often just a few percent of the customer's revenue.
4. A specific call to action: a specific time, a minimal step: "Do you have 15 minutes? I can call at 3:30 this afternoon or before noon tomorrow." Every needless back-and-forth leaks about 5% of the opportunity, and a vague invitation like "what do you think?" is drilling a hole in your own boat.

Part 4: The offer formula: conversion = ROI times trust, divided by friction

The tightest and sharpest piece of thinking in the course: the closing rate of an offer = (value the customer believes they will get) times (the trust you can deliver) divided by (the hassle of getting started). Three dials: ROI must be quantified and time-bound ("20 meetings in 60 days" rather than "lots of meetings"); trust is built with the four levers you have learned (proof, authority, similarity, identity); friction is reduced by stating it only takes a few minutes, with a ready-made way in, and by reversing the risk onto yourself (no results, no payment). The curriculum cites a stack of real offers that made money in exactly this pattern: a commitment to a number of meetings in 60 days or no fee; building a free sample product within 48 hours, and working together only if the customer likes it; writing the first 500 words for free. The common thread is easy to see: the seller carries the risk for the buyer: it hurts the wallet in the short term and wins trust in the long term. And one distinction worth engraving: understand the system rather than memorize the templates: no matter how good a template is, it wears out when everyone uses it; grasp the principle and you can generate new templates all year.

Part 5: Dissecting real emails: recurring mistakes and how to fix them

Nearly half the course is spent "dissecting" a dozen real emails received in the wild, scoring them against the seven levers and rewriting them. The classic mistakes are gathered in one place so you can check your own letters:

• Exposed personalization variables: a channel name scanned as a person's name, a video title dropped in stiff quotation marks mid-sentence: the recipient knows instantly it is a mail merge. Fix: variables must flow as if typed by hand, and often the cheapest approach (taking the first word of the display name) beats a fancy AI.
• Signatures and tool traces: a line saying "sent with software X", a grand footer like a press release: the mass-mail smell is overpowering.
• No links in the first email: spam filters treat links from strangers as a scam signal, and so do recipients. If you want to send material, ask permission first ("can I send 2-3 samples?"): it both keeps you clear of filters and earns you another small commitment.
• Weak bragging beats no bragging: "I design cover images" is the weakest proof; "I am making cover images for a few channels with a combined one or two million followers" is real proof. If you have no big number, use a small one that is true, and never make one up.
• Machine-speak (LLM-isms): a letter that calls itself "I am an AI running 10 agents" or is full of empty flowery phrases scores near zero on all seven levers. Simple, specific language with numbers always wins.
• The offer buried at the end: a long letter is not wrong, but the main dish must be served early: readers drop off line by line.

Part 6: Subject lines, follow-ups, and tailoring for each channel

• Subject lines: the "can't conclude yet" principle: a subject line should not sell, only spark enough curiosity that the email has to be opened: the recipient's name in lowercase, "Nam's blind spot", "from the way you started". A subject line that plays on fear of loss ("you are wasting this much every month") beats one promising gain: people react to loss more strongly than to gain.
• Follow-ups: two is plenty. The standard starter set is just the original email plus one short, natural follow-up in a real person's voice ("I'm messaging again in case you missed it"). Add a third or fourth only once the campaign has proven a good reply rate: each extra email adds a bit of response but also a bit more risk of being flagged as spam.
• Tailoring for each channel: the same campaign must be "reshaped" for each platform, not copied verbatim. Email: readers see the sender's name first, then the subject, then the preview line: the first 150 characters decide its fate. LinkedIn: a profile photo with a contrasting background, with your title and partner certifications pulled to the top. Social networks like X and Instagram: most messages land in a requests inbox, and the tone is far more casual. SMS: the preview line is nearly the whole message, so say everything in ninety characters. In Vietnam, swap those channels for Zalo and Facebook: the principles stay the same: the first line decides, the tone fits the channel, and the first message absolutely has no link.

Part 7: Where AI fits in this craft: a contrarian view worth respecting

This is the part we want you to read slowest, because it runs against every pitch out there: an automation curriculum, taught by someone who lives off automation, declares that they rarely use AI to write sales letters. The reason comes with real data from the curriculum's own community of thousands of practitioners: AI takes a newcomer to 70-80% of the skill very quickly, but a 70-80 point letter now falls below the passing line: the whole market has AI, so the "read it and delete it" bar keeps rising, and the group that relies on AI to write for them gets worse results than the group that writes systematically themselves. Where AI that are is used, and used wisely: filling small variables into a fixed template (picking each customer's "most notable asset" to fit into a genuine compliment); a "humanizing" layer (shortening company names and place names to how insiders say them, creating a real sense of familiarity); and scanning and enriching data: exactly the machines you built in the n8n and agentic courses. In other words: the machine does the muscle work, the human keeps the voice: it should sound familiar, because it is the philosophy we have repeated throughout the series, and how we run content at MONA ourselves.

The course also lists the "gray" techniques spreading out there: buying pre-warmed accounts, mass calling with parallel auto-dialers, faking iPhone text messages: listed so you can recognize and avoid them, because the price is accounts that vanish and legal risk, not to mention that in Vietnam the decree against spam messages is no joke. Our principle: decent cold outreach is the right person, few but good, give before you ask: that road is a few weeks slower but you can walk it for years.

Part 8: Closing the 12-course series: the machines are enough, the voice is yours

That completes the set: from using AI for work, building your own software, setting up self-running workflows, through to launching a service and the opening sales letter. Looking back, you will see a thread running through it that we kept on purpose: machines amplify, humans steer: a machine can scan a thousand prospects but the greeting has to carry a human voice; a machine can build an app in one session but the safety boundaries are drawn by people; a machine can run a workflow all night but the signature that takes responsibility is a person's. The businesses that win in the AI era, from what we have seen after over 10 years and 14,000+ projects, are those that learn the machine part fast and hold the human part firmly.

A simple closing invitation to the series: if you want to apply these courses to your company and want a team alongside you, from roadmap consulting and building systems to training your staff, call 1900 636 648 or send it to info@themona.global. MONA listens first, gives a free assessment and a transparent quote, and what we sign is the price we deliver. And if your problem does not need us yet, we will say so plainly, just as we have said things plainly through all twelve courses.

Quick answers

Does cold email still work? Yes, provided you do it scientifically: real personalization, an offer with a commitment, and enough samples in your tests. The campaigns in the curriculum reached 4-15% reply rates, while mass emails are happy with 1%.

How many emails a day is safe? A few dozen per mailbox per day, spread across natural hours, with a plain-text first email and no links, and a sending domain separate from your main domain. Raising volume before quality is secure is how you burn a domain.

Should AI write sales emails? Not the whole letter. Use AI for the muscle work: scanning data, filling variables in a template, shortening names so they sound natural; the template and the voice should be human. Data from a community of thousands of practitioners shows that the group relying on AI to write for them gets worse results.

How does this apply to Zalo and Facebook in Vietnam? The principles stay the same: the first line decides, tone fits the channel, no links in the first message, give before you ask. Swap "email" for "message" and the subject line for an opening line: the seven levers and the four-step formula work as they are.

What tools do I need to start? A dedicated sending tool with A/B testing, a source of prospect data, and the personalization engine you learned in the n8n or agentic workflow courses. But the right order is: hand-write your first 50 emails to find your voice, then plug in the amplifying machine.

MONA's free AI course series pauses here. The full list of 12 courses is in the AI coursessection, and new courses will keep being added. Wherever you get stuck as you learn, just call us.