I endured 700+ cold marketing emails to research why they don’t work

Cold outreach emails suck. There, I said it.

Unfortunately, cold emails are also ubiquitous. Since I started working at a marketing agency five years ago, I’ve endured a daily onslaught of unasked-for emails—offers for leads, new software demos, and website backlinks.

Normally, I immediately delete these emails or mark them as spam. They clutter my inbox, constantly clamoring to book a demo or grab me for a quick chat. It’s annoying and the antithesis of my view of effective marketing.

Then, at the start of 2025, I thought of a better use for these cold emails—research. What if I collected these emails and studied their ways? What if I could expose the lazy patterns of cold email blasting and statistically demonstrate how ineffective (and irritating) this practice is?

My methodology

I was remiss in having deleted all these spam emails for the last five years—all that juicy data lost. However, I receive such a volume of cold emails that it didn’t take long to collect a new sample of data to study. In less than six months, I had over 700 emails to study.

My last big marketing research experiment was done entirely manually. It took me hours to compile and analyze one piece at a time. To handle this volume, I turned to a new resource: LLMs. I compiled all these emails and outsourced the analysis to ChatGPT 4o.

I created a new folder in Gmail to categorize the emails. Every time a new cold email hit my inbox, I tagged it. ChatGPT did the rest, with my guidance. I asked AI to write a custom script to build a database from this Gmail folder. It scraped the data from each email to collect certain fields:

• Send date and time
• Subject line
• Sender’s name and email address
• Body of the email and word count
• Number of emails from the same sender

Then, I worked with the LLM to build a few more subjective fields like:

• Tone of the email
• Service offering
• If they included a link, a P.S., or a call to action
• If they mentioned my name or company name

My Google Sheets script automatically scraped the folder daily and updated the database. This gave me an instant snapshot of what the emails contained. Because this was done automatically and using AI, I could add new data points to reference and re-run the analysis instantly.

The rules

These are the rules I followed to be as objective and fair as possible during this process:

• Only include emails that come directly to my email
• Open every email to read the contents
• Don’t click on any links or reply to any emails
• No generic email newsletters—only direct sales emails
• No clearly scam emails
• Can’t submit any lead gen forms to trigger an email, because then it’s not cold outreach

My goal

I’ll admit it, I’m already biased against cold emails. So I never expected to assess whether or not they work. Data shows that cold emails sort of work.

Studies vary in results: from a 30-50% response rate (doubtful) to closer to 1%. Even articles discussing how to do “better” cold emailing admit that “No one wants to send them, no one wants to receive them.” The problem is that most of these studies are done by companies that offer cold outreach services. They’re measuring by sending emails, not receiving them.

At scale, you’ll get a few responses and maybe a few new sales. However, this means you have to harass (and risk pissing off) countless more people. How many angry potential customers is one new customer worth?

Cold emails have yet to work on me. At best, I ignore them. At worst, I mark them as spam and then write an entire research article exposing their practices.

My deeper question is what approaches do cold outreach salespeople think work. We’ll call it their “worst practices” (because “best practices” is too generous). So I set out to collect as many cold emails as possible and identify behavior patterns and content cliches.

The data

• 756 emails collected
• 180 days
• 4.3 emails per day
• 332 senders

I wanted to have enough data to be statistically significant. With ChatGPT’s input, I determined that I needed at least 663 emails to reach a 99% confidence level with a 5% margin of error. By collecting even more than that, we can confidently say this research means something.

You can check out the entire database yourself at this link. (I will keep collecting data in real time even after this article is published, so the numbers may change slightly over time.)

8 Lessons from the data

1. Multiple emails more than half the time

54% of senders reached out more than once. The average number of outreaches in a single thread was almost three emails. The most emails from a single person was 15.

Some of the common refrains in these constant follow-ups included references to:

• My last/previous email (59 times)
• My last/previous message (37x)
• Getting in touch with someone else at the company (30x)
• Dropping into my inbox (17x)
• Not getting buried under other emails (16x)

They won’t take no (or a cold shoulder) as an answer.

2. Almost all emails are predictably “personalized”

84% of emails were personalized with my name. 45% of the emails mentioned my agency name (GreenMellen) at some point. There are also a handful of times I’ve been called the wrong name (like George) or by the name of the person sending the email (I’m looking at you, Joseph Wilson).

There were also plenty of random mentions of my job title, projects on our website, or other meaningless pieces of information from my LinkedIn profile (“Congrats on that online certification you earned six years ago!”). Plenty of the personal details were just wrong—like the email congratulating me on my first work anniversary when I’d been at the company for over 4.5 years.

3. Emails are generally short, but rarely sweet

Even (most) cold emailers know they don’t have your attention for long. The average word count for the emails was just over 180 words. The longest was 1,440 words; the shortest only 18.

The automated script also included a score for readability using the Flesch Reading Ease score. It’s a scale from 0-100, with a higher score being better. The average readability of all the emails was 47, which Flesch ranks as “very difficult” to read and equates to a 10-12 grade reading level.

I also fed the text (more than 125,000 words, or the length of a long fiction book) into a word cloud generator to detect the most-used words.

4. They use confusing subject lines to act like clickbait

Subject lines are important to email open rates. As such, cold emails overly like to personalize them—35% of the subject lines included my name.

The average length of the email subject lines was 5 words. The longest subject line was a whopping 40 words long; and 22 of the emails had a subject line of only a single word.

Just like with the email bodies, I extracted all the text and generated a word cloud.

Oh, and here’s a list of some of my favorite, terrible subject lines:

• 60 minutes each day. Yours again
• Want To Put This On Ice?
• one thing Robert
• Lets Try That Again
• Robert, can you handle more sales?
• More on your quiz: What Kind of Mom Are You
• GreenMellen?
• You Deserve Better
• it’s honestly absurd Robert

5. Their goal is to get you to respond, not click

Only about 37% of emails included a link. Usually, that link was to book a calendar meeting with them, but it could also include examples of their work or backlinks they could sell me.

Only 13% included a clear call to action, like booking a meeting. Presumably, their main goal is for you to respond to the email (without them mentioning it specifically—like a sort of marketing inception) to show interest. That’s how they hook you.

Speaking of links, only 19% of the emails mentioned an unsubscribe link. Technically, you’re supposed to include one, but since I never subscribed to receive their emails, they don’t have to let me unsubscribe.

6. Over half of the emails are sent in the morning

More than half (57%) of the cold emails are sent in the morning—between 6 am and noon. Another third (35%) was sent in the afternoon, from noon to 6 pm. A small fraction was sent outside of that time—either the very early morning or late evening. The hour with the most emails was 10-11 am.

This shouldn’t come as a surprise. Most emails are sent in the morning, and an even larger percentage at some point during the work day. Most people aren’t checking emails regularly outside of business hours.

7. We have different definitions of “value”

As an excuse to keep emailing me, they invent many reasons to follow up. Besides the empty ones like raising their email to the top of my inbox, there were quite a few “value-driven” emails—at least valuable to them.

These are some of the most common refrains from these emails (and how many emails each phrase was mentioned in):

• A quick chat (145) on Zoom (43)
• A guarantee (36) of ROI (66)
• Offers a gift card (26)—usually Amazon (14)
• Offers a demo (38)
• Free trial (6) with no obligation (16)
• Personalized Loom video (22)
• Some sort of free download (11)

8. Using cold emails to sell… more cold emails

With so many emails, it’s easy to wonder what services and products these people are actually selling. What are they pushing so hard? Remember I work for a marketing agency, because this certainly shapes the kind of offers I’m receiving.

I did a quick check to see how often some of these terms were mentioned in the emails:

• Ads: 149
• Leads: 141
• AI: 106
• Video: 90
• SaaS: 83
• SEO: 66
• Partnership: 27
• Backlinks: 18
• Sponsorship: 16
• Graphic design: 8

Most of the lead generation services themselves are selling more cold outreach. I guess this is mainly just people using their own tools to sell the tool. 41 emails specifically mentioned “cold emails” directly—talk about meta.

The other irony here is that we were often offered services we provide to our clients. There were a handful of offers for website development, graphic design, and SEO—all are services we sell. Nothing more tone deaf than offering to improve the website of a website agency.

2 Reasons why you shouldn’t send cold emails

People will still send cold emails—no matter how many articles I write or cold emails we ignore. For many reasons, but primarily because they can. There’s nothing legal stopping them.

My main argument against cold emails is that it’s entirely unethical. It gives marketing a bad name and makes even legitimate channels more difficult to reach people. Every cold email devalues trust in marketing—even for those who don’t practice it. But if the ethics argument worked, no one would be sending cold emails. We wouldn’t need to have this conversation.

Instead, I’ll appeal to their baser nature: results. Even if you send the “best” cold email, is it really effective when you’re playing in a toxic environment? If I get 5 cold emails per day, I’m not going to respond to the one good one—I’m going to ignore all of them.

So please stop trying to send “better” cold outreach because everyone around you is just optimizing for scale, and quickly degrading quality. When you play that game, everyone loses.

3 pieces of advice for avoiding cold emails

Don’t engage

Opening, clicking, or responding is their signal that you’re interested. Once you target yourself, that only opens the floodgates even more. Your best bet is to treat them like hypothetical bullies—ignore them and hope they go away.

Mark them as spam

Spamming people is legal, according to the U.S. CAN-SPAM Act. Your best protection against unwanted emails is your email service provider (ESP). Labeling these incoming emails as spam lets Gmail, Outlook, etc. know that these senders are shady and unwanted, not just for you, but for others, too.

Don’t send cold emails yourself

Here’s the golden rule of marketing—don’t participate in any marketing activity you wouldn’t want to be targeted with yourself. So, unless you enjoy being interrupted all day with sales emails, then don’t inflict the same treatment on others. What should you do instead?

• Use permission-based marketing to collect contacts
• Send out emails to a list you’ve legitimately collected with content people actually want
• Meet your target audience in person at industry events or conferences
• Join online communities like Reddit or Facebook groups to build a relationship first
• Create value-based content that builds trust and curiosity with the right people
• Get creative to find methods other people aren’t already using (and abusing)

Where do we go from here?

I’ve been working on this article for over six months—carefully collecting emails, observing patterns, and crafting this passionate appeal against cold email. Who knows if it makes a difference? But at least I have real data to point to on why people should stop.

In the meantime, I’ll probably continue to collect these emails. Maybe I’ll revisit this topic again in another 2-3 years after I’ve collected another few thousand examples. I know it’ll be a success when some cold emailer casually mentions how excited he was to read my awesome new article on cold emails… only to try to sell me more leads I don’t want.

In the meantime, you can also play Cold Email Bingo with me.