Gains:
- Ability to produce cold emails based on trigger events that do not go to spam
- Ability to personalize LinkedIn connections and message threads
- Ability to open a cold call and prepare short scripts and use them flexibly
Cold outreach is the most difficult but most misunderstood area of sales. Most salespeople think "if I send more, I will get more responses" and send thousands of generic e-mails; The result is a spam folder, a low response rate, and an email domain that becomes discredited over time. However, good cold outreach is not a volume game, but a relevance game. In this unit, you will learn how to use artificial intelligence correctly to create short, personalized messages based on a real triggering event; You'll learn how to flexibly design LinkedIn and phone scenarios. Our goal is not more, but more accurate communication.
Anatomy of Cold Email
The job of a good cold email is not to persuade the other party; His job is simply to invite you to the next step (a short conversation, a question, an answer). Emails that achieve this have four features in common.
feature
What do you mean?
Why is it important?
brevity
50-125 words, 3-4 lines on mobile
The decision maker opens dozens of emails a day; Long emails are deleted without being read
Single CTA
Just one clear request
An e-mail asking two questions does not receive a response to any of them.
Triggering event
Based on a real signal
Answers the question "Why now, why me?"
Personalization
First sentence specific to the company/person
Proves there is no bulk shipping
Trigger event is the key concept here. A new manager appointment, a job posting, an investment round received, a new product launch, a legislative change or a content published by the company; All of these provide the justification that "it makes sense to talk about this right now." AI can't invent these signals on its own, but it is very good at turning the signal you come up with into a powerful opening line.
Deliverability: Why Bulk Shipping Sinks
Deliverability is whether your email lands in the inbox or spam folder. When you copy and send the same template to hundreds of people, email providers recognize patterns: same body, same link, sudden increase in volume. These trigger spam filters. Additionally, sending to a large number of invalid addresses (high bounce rate) reduces the reputation of your domain name. Result: even the emails you sent to valid customers no longer arrive. So even AI-generated content should be personalized, the submission volume should be kept reasonable, and the list should be clean. Artificial intelligence helps you generate variations; It breaks the urge to send the exact same text to every recipient.
Cold Email Generation with Artificial Intelligence
If you give good input to the AI, you will get good drafts. In the prompts below, fill in the fields in {{ }} with your own data.
Role: You are an experienced SDR in B2B SaaS sales. Task: Write a SHORT cold email draft based on the following information.To: {{ name, title }}, {{ company }}Triggering event: {{ e.g. they opened 15 sales representative ads last month }}Our solution: {{ what are we doing in a sentence }}Targeted value: {{ e.g. shortening ramp time for new hires }}Rules:- 90 words maximum.- One call: Offer a 15-minute meeting.- Let the first sentence address the triggering event, no generic praise.- Don't use exaggerated promises, guarantees, or clichés like "industry leader".- Let there be one question at the end. Generate 3 alternatives for the subject line.
A separate round for the subject line works:
Suggest 5 subject lines for the following email body: {{ body }}Restrictions: No more than 4 words, no clickbait, no shouting in all caps, no emojis. Make it intriguing but be honest.
Weak Prompt / Strong Prompt
Weak prompt:
Write a sales email for a software company.
This prompt produces a generic, trigger-free, “we have a custom solution for you” style text that is probably 200 words long; So it's direct spam.
Powerful prompt:
Write a cold email to {{ Name }}, HR Director for {{ Company }}.Trigger: This week he posted on LinkedIn that "we're re-engineering our onboarding process."We: a platform that provides interactive micro-training to new employees in their first 30 days. Typical impact we measured: significant reduction in ramp time (varies by customer, not guaranteed).80 words, single CTA (short call), friendly but professional tone. Use measurable but understated language.
The difference is in the constraints of specificity and honesty. The second prompt produces a text that makes the other party feel like "this has read my situation."
Tip: Always rewrite or manually proofread the first sentence of the AI-generated draft. The receiver just looks at that sentence and decides whether to continue or not; Put the most human touch there.
LinkedIn: Connection and Message Thread
A non-aggressive approach on LinkedIn is even more important than email. Turning a connection request into a sales pitch is the most common mistake. Recommended rhythm:
- Connection request: A short, non-salesy note (or sometimes no note) that touches on common context.
- Post-acceptance thanks: Don't market the product right away; Share a relevant observation or question.
- Share value: Send useful content, data, or a small insight.
- Soft invitation: Offer a meeting only after you see a signal of interest.
If you're using InMail (a paid message to people you don't have a connection to), the subject line and first line are even more critical because the recipient doesn't know you.
Write a LinkedIn connection request note (max 250 characters).Person: {{ name, title, company }}Common context: {{ e.g. He spoke at the same industry event / I liked his post }}Rule: Do not sell products, do not use the cliché "I am growing my network". Let it be clear and sincere why I want to connect.
My connection has been accepted. The person is {{ name }}, {{ role }} at {{ company }}. Write a helpful SHORT message that I can send before moving on to sales. Include an observation + an open-ended question. Don't ask for an appointment.
Caution: Sending an automated sales message as soon as the connection is accepted is the fastest way to get blocked (and burn brand reputation) on LinkedIn. Sending bulk messages through automation tools may also violate platform rules. Focus on making the relationship real, not on speeding it up.
Cold Call: Flexible Scenario
The phone is still effective, but reading the script word for word is a disaster. Think of the scenario as a guide, a map. Parts of a good opening:
- Pattern interrupt: Breaking the expected "hello, how are you, I want to introduce you to a product" pattern. That's what honest self-introduction does.
- Asking for permission: "What if I disturb you for exactly 27 seconds, and then you decide whether to continue or not?" A sentence like this gives control to the other party and brings down the wall.
- Value sentence: State the problem you solved, not the product, in one sentence.
Write a flexible opening script for a B2B cold call. Product: {{ tagline }}Target persona: {{ role }}Include: (1) an honest pattern interrupt, (2) a permission-seeking statement, (3) a one-sentence value proposition, (4) an open-ended exploratory question. Also include a polite, no-pressure response for “I'm not available right now” and “I'm not interested.” Write humanly, not robotically.
Mini Case: Cloud Accounting Software
Zeynep is an SDR at a company that sells cloud accounting software to SMEs. "Aydın Logistics" is on the target list. On LinkedIn he sees that the company opened its second location last week (triggering event). It gives artificial intelligence this information and the challenge of consolidated reporting in a company's multi-branch businesses. AI produces a 75-word draft that begins, "I saw your second branch opening, congratulations; the most common challenge we hear in multi-branch businesses is seeing branch-based financials on a single screen." Zeynep manually simplifies the first sentence a little and suggests a 15-minute meeting as a single CTA. The answer comes the same day: "Actually, that's exactly what we were talking about." The difference was not in volume but in the right observation at the right moment.
Common Mistakes
- Bulk sending the same text: Kills Deliverability, reduces it to spam. Differentiate each message with at least one factual detail.
- Two or more CTAs: “Look at this document and let's schedule a meeting” produces no response.
- Exaggerated promise: "We will increase your sales by 300%" is unverifiable and destroys trust. Use spacing and “it depends on the customer” language.
- Starting with the product: Explaining your features in the first sentence. First their world, then your solution.
- Sending AI output as is: AI can sometimes make up a “success story” or number that doesn’t exist. Verify each statistic before submitting it.
- Not following up: Most replies do not come in the first email, but in 2-3 polite follow-up emails; But follow-up should be a reminder, not an insistence.
In summary
- Cold outreach is a game of relevance, not volume; brevity, single CTA, triggering event, and personalization are the four cornerstones.
- To maintain deliverability, avoid bulk generic shipping, keep the list clean, differentiate each message.
- Use AI to generate drafts and variations; manually verify the first sentence and all figures.
- On LinkedIn, relationship first, sales second; Avoid automated sales messages.
- In cold calling, use the script as a flexible map: pattern interrupt, permission request, one-sentence value.
- Understated, verifiable and unpressured language is both ethical and more effective.
Application task
Choose a real target customer (either from your existing list or a fictitious realistic company). First, find a real triggering event for that company (news, announcement, share). Then, use the powerful prompt pattern in this unit to have the artificial intelligence produce (a) an 80-word cold email, (b) a LinkedIn connection note, (c) a cold call opening script. Rewrite the first sentences of the deliverables yourself, verify all the numbers, and make sure there is a single CTA in each. Finally, put it side by side with a weak version and note the difference in one sentence.