How We Personalize 1,200 Cold Emails a Day with a 3-Step Automation

AI Automation

Our cold emails were landing in spam. We built a 90-minute Make.com automation that shortens company names and generates personalized icebreakers for 1,200+ emails a day.

Our cold emails were landing in spam

We send between 900 and 1,500 cold emails a day. Monday through Saturday. At that volume, personalization was never an option. Every message went out with the same template, the same opening line, the same forgettable intro.

The inbox placement tests told the story. Emails were landing in spam folders. Prospects never saw them. The ones that did get through looked exactly like what they were: mass outreach from a stranger.

INFO: The math we were ignoring Manually personalizing each email takes about 8 minutes. LinkedIn research, company context, writing the icebreaker. At 1,200 emails a day, that is 160 hours of research. Every single day. We needed a third option.

We were stuck between two bad options. Send garbage at scale, or send quality to nobody.

So we built a system that does both.

A 3-step automation built in 90 minutes

The entire system is three modules in Make.com. It reads from a Google Sheet, uses GPT to shorten the company name and generate a personalized icebreaker, and writes it back. That is it.


Step 1: Pull leads from Google Sheets

The scenario runs on a schedule, pulling every lead that does not have an icebreaker yet. Your sheet needs two things at minimum: a business name and a one-liner about what they do.

Optional but helpful: LinkedIn headline, job title, company size, industry, recent posts, tech stack. Each additional data point makes the icebreaker sharper.

Step 2: Shorten the name and generate the icebreaker

This is where it gets interesting. GPT does two things in one pass.

First, it shortens the company name. If someone works at "TechFlow Solutions LLC," the system just calls them "TechFlow." If it is "Anderson & Partners Consulting Group," it becomes "Anderson."

Why? Because that is how you would actually say it in conversation. Nobody emails a friend and says "I was looking at Anderson & Partners Consulting Group." You say "I was looking at Anderson."

That one small change makes the entire email feel less corporate. Less mass-produced. More like someone who actually knows who they are writing to.

Second, it generates a one-liner icebreaker using whatever data you have on the lead. Their LinkedIn headline, their company description, their industry. The prompt tells GPT to reference something specific and connect it naturally to why you are reaching out.

TIP: The formula The icebreaker references something from their profile (a role, a stated priority, a company detail) and bridges to your value prop. It should read like you spent 5 minutes researching them. The AI does it in under a second.


Step 3: Write it back

The generated icebreaker and shortened company name write back to the same Google Sheet row. Your sending platform picks it up from there. No exports, no imports, no copy-pasting.

No new tools to learn. No new dashboards. The icebreaker just appears in the column next to the lead.

• Make.com: Free tier available
• Google Sheets
• OpenAI GPT
• Apollo.io
• Instantly

What changed

• 1,200+: Emails Personalized/Day
• 90 min: Total Build Time
• ~$0: Per Email Cost
• 6 days/wk: Runs Automatically

Before this automation, every cold email was identical. The same opener, the same pitch, the same result: spam folder or delete.

After: every email references something specific about the prospect. Their shortened company name (like a friend would use it, not a formal pitch), their role, their situation. It reads like someone actually looked them up before hitting send. Because technically, something did.

Deliverability

The inbox placement problem largely went away. Personalized emails look different to spam filters. Unique content in each message, specific references, natural language. Mailbox providers stopped flagging us.

Response rates

People reply when they feel seen. An email that mentions their actual company by its shortened name and references their actual role gets read differently than "Hi {first_name}, I noticed your company is growing."

The response rate went up because the emails stopped looking like mass outreach.

The time equivalent

If we had done this manually at 8 minutes per email across 1,200 emails a day, that is 160 hours of research. Every single day. The automation handles it in the background while we do actual work.

SUCCESS: The real win This is not about saving time on something we were already doing. We were not personalizing emails before because it was impossible at our volume. The automation made personalization possible for the first time. That is the difference.

Before and after

"Hi {first_name}, I came across your profile and thought we could connect. We help companies like yours improve their marketing with AI automation. Would you be open to a quick chat?" — Before: Generic Template, Same message to every prospect

"Hi Sarah, saw TechFlow just closed the Series B. Hiring 4 SDRs is one way to hit that outbound target. We built a system for a similar stage SaaS company that personalized 1,200 emails a day without adding headcount. Happy to share the playbook if useful." — After: AI-Generated Icebreaker, References her LinkedIn activity, company stage, and hiring signals

Same prospect. Two completely different emails. The first gets deleted. The second gets a reply.

Who this works for

Anyone sending cold emails who has hit the wall between quality and volume:

• **B2B sales teams **sending 50 to 5,000 emails a day
• **Agencies **running outbound for multiple clients
• **Recruiters **reaching out to passive candidates
• **SaaS founders **doing their own sales outreach
• **Consultants **building a pipeline without a sales team

If you have a lead list in a spreadsheet and you are sending the same email to everyone on it, this system fixes that in 90 minutes.