5 Cold Email Mistakes That Are Killing Your Results (From 18 Million Emails Sent)
Troy Aitken shares 10 years of cold email lessons: deliverability, data sourcing, AI use, and A/B testing that actually books meetings.
I've sent over 18 million cold emails in the past two years. Before that, I spent a decade testing, breaking, and rebuilding cold email systems across hundreds of niches. And the same five mistakes keep showing up, over and over, in almost every business I talk to. Fix these and you'll outperform 90% of the market. Keep making them and it doesn't matter how good your offer is.
1. Deliverability Is the Foundation, and Most People Are Getting It Wrong
What worked five years ago doesn't work now. Back then, you could buy a handful of domains from Namecheap or GoDaddy, spin up sending accounts, run a two-week warmup, and be off to the races. That era is over.
Google and Microsoft have tightened restrictions because everyone and their mother is now running cold email. The irony is that cold email is actually more effective than ever, but only if you take a deliberate approach to deliverability that most people skip.
Here's what we do at the infrastructure level:
50/50 mix of G Suite and Outlook accounts. This lets you identify which provider is resonating and which isn't before you scale.
Private IP parking. When you buy a domain from Namecheap, GoDaddy, or Squarespace, it sits on their shared IP. If other users on that IP are spamming, you get penalized alongside them. Move your domain to a privately owned IP address.
Automated inbox placement tests. Run these consistently, not just at setup.
50/50 split of warmed and non-warmed inboxes. Tools like Lemlist, Instantly, and Smartlead use shared warmup ecosystems where fake emails bounce back and forth to simulate engagement. We test whether that warmup is helping, hurting, or doing nothing, and adjust strategy accordingly.
Do these four things and you're already outworking most of the competition on infrastructure alone.
2. The Data That's Hardest to Find Yields the Highest ROI
Most businesses go straight to Apollo, pull every contact they can, and start blasting. I understand why, it's cheap and fast. But that's exactly the problem.
Apollo is everyone's first stop. The people in that database are already getting hammered with cold emails from companies just like yours. You're not entering a market, you're entering a queue.
The businesses that see outsized results are the ones willing to find data nobody else is using. A few methods we've used directly:
PhantomBuster to scrape LinkedIn groups and pages. You can pull company and contact info, then call out specifically how you found them in your outreach. That specificity alone lifts reply rates.
Custom Google scrapers targeting Google Maps, Google My Business, and Google Search. We then run those contacts through Clay for enrichment.
Clay's waterfall enrichment across six to seven data providers. Apollo is one of them, but only one. Running the same target persona through multiple providers increases the chance of finding accurate contact info and reduces the overlap with every other cold emailer using a single-source list.
The harder the data is to find, the less competition you have in the inbox.
3. Templates Are a Starting Point, Not a Strategy
Cold email templates are everywhere. They're fine as a reference if you've never written a cold email before. But copying and pasting one without understanding why it's structured the way it is will get you nowhere.
When I look at a template, I'm asking: what is the actual persuasion strategy here? Is it painting a picture of a future state? Calling out a specific pain point? Positioning the sender as a familiar peer? Each of those is a different approach, and each one works differently depending on the audience.
The right way to use a template is to reverse-engineer the strategy, then rebuild it around your specific offer, your specific audience, and the data you just worked hard to find. A template written for a SaaS company selling to HR directors isn't going to land the same way for a professional services firm targeting operations leads. Understand the structure, then make it yours.
📥 Best Email Warmup Tools
The 6 warmup tools that work — ranked by an agency managing 20,000+ inboxes.
4. AI Should Inform Your Copy, Not Write It
AI didn't make cold email easier. It made lazy cold email more common, which made good cold email more valuable.
Nine times out of ten, when someone hands their entire email sequence to ChatGPT, the output is detectable. Recipients can feel it. And there's a reasonable argument that Google and Microsoft can too.
That said, I use AI constantly, just not as a ghostwriter. Here's where it actually adds value:
Market research on your ICP. Ask it what keeps your ideal client up at night, what their aspirations are, what obstacles they hit daily. Use that to write more targeted copy yourself.
Generating content angles. We use AI to brainstorm ideas for things like SEO content, website audits, or marketing insights we can reference in outreach to show we've done our homework.
Tightening copy you've already written. Write the email yourself, then paste it into GPT and ask it to make the language more concise or precise. You'll get a few variations you can merge or A/B test.
Verifying email accuracy. A practical, low-risk use that saves time without compromising quality.
The rule I follow: AI informs the process, a human writes the email.
5. A/B Testing Doesn't Stop at Subject Lines
This is the one practice I credit most directly with generating over $8 million in revenue for us and our clients. And most people are only doing half of it.
A few years ago, we worked with a client named Bob from Life360. He had a B2C program he wanted to take B2B by partnering with other businesses. He came to us with four industries he wanted to target. We added three more we thought had potential. Over 45 days, we tested all seven and narrowed it down to three that were booking meetings consistently and converting well.
That's the point: A/B testing applies to your target populations, not just your copy.
On the copy side, we test subject lines, intro strategies, offers, positioning, and calls to action. For a first campaign, most of our clients run 8 to 10 variants of step one alone. We'll hold the offer and CTA constant while rotating three different intros, then hold the intro constant and rotate three different offers, and so on. After roughly 400 sends, you have clear winners, or you have clear knowledge of what doesn't work, which is equally valuable.
Keep testing new angles. The businesses that win at cold email long-term are the ones that never stop.
Key Takeaways
Deliverability requires private IP parking, a G Suite/Outlook mix, automated inbox placement tests, and monitored warmup ratios.
Apollo-only data strategies put you in the same inbox as every competitor. Scraping LinkedIn with PhantomBuster, using Google-based scrapers, and running Clay's waterfall enrichment across multiple providers gives you an edge.
Templates teach structure. They don't replace understanding the persuasion strategy behind the copy.
Use AI for research, angle generation, and tightening your own drafts. Don't use it to write sequences from scratch.
A/B test your target industries and audience segments, not just your email copy. After 400 sends, you'll have data worth acting on.
Frequently Asked Questions
Why does parking a domain on a private IP address improve deliverability? When you buy a domain from a registrar like Namecheap or GoDaddy, it sits on a shared IP address alongside thousands of other users. If any of those users are sending spam, your domain's reputation gets pulled down with theirs. Moving to a privately owned IP address isolates your sending reputation so you're only accountable for your own activity.
What's wrong with using Apollo as your only data source? Apollo is the most commonly used prospecting tool in cold email, which means the contacts in its database are already receiving a high volume of outreach from competitors in your space. Sourcing data through LinkedIn scraping, Google Maps, or Clay's multi-provider waterfall enrichment gives you access to contacts who aren't already saturated with similar pitches.
How many A/B test variants should a first cold email campaign run? Most of our clients start with 8 to 10 variants of their first step. The structure is to hold two elements constant (say, the offer and the CTA) while rotating three different intros, then flip it and hold the intro constant while testing three different offers. After roughly 400 sends per variant group, patterns become clear enough to scale the winners.
What's the right way to use AI in cold email without it hurting results? Use AI to conduct research on your ideal client persona, generate content angles to reference in outreach, and tighten copy you've already written yourself. Avoid having it write full email sequences from scratch. AI-generated emails are detectable by recipients and likely by spam filters too. The human-written draft informed by AI research consistently outperforms the fully AI-generated one.
Your pipeline, rebuilt.
20-minute strategy call. We'll audit your ICP, show you which signals we'd track, and map out exactly what the first 120 days would look like. No commitment, no pressure, no pitch deck.
