5 Cold Email Lessons From Scaling to $6.5M ARR With Instantly

We scaled ColdIQ past $6.5M in ARR with Instantly powering roughly 70% of our cold email sending. Millions of emails. Thousands of campaigns. Hundreds of clients across every B2B vertical you can think of.

Along the way, we made every mistake in the book. We burned domains, wrote emails nobody replied to, and watched good leads go cold because we were too slow to follow up.

These are the five lessons that made the biggest difference. Not theory. Not "best practices" pulled from a blog post. These are patterns we learned by running real campaigns, tracking the data, and iterating until something worked.

1. Lead With Value, Not Asks

The default cold email approach is to ask for something. "Can I get 15 minutes of your time?" "Can I pick your brain on outbound?" "Would you be open to a quick call?"

Every one of those puts the burden on the recipient. You are asking a stranger to give you their time with zero evidence that you can deliver value. The math does not work in your favor.

We flipped this completely. Instead of asking, we give.

The framework is simple. Before writing a single email, we ask ourselves: if this person signed as a client tomorrow, what is the first thing we would do for them? Then we do that thing. For free. In the email.

Here is a real example. We ran a campaign targeting recruiting firms. The first thing we would do for any recruiting firm that hired us is build a list of HR Directors at organizations hiring for 10+ roles. So we built the list. We sent it to them. And we attached a short tutorial showing how they could build the same list themselves using Clay and Apollo.

The result was more inbound interest than we could handle. Recruiting firms forwarded the list to colleagues. Some replied just to say thanks. Others booked calls because if the free version was this good, they wanted to see what the paid version looked like.

This works because you are demonstrating competence, not claiming it. You are showing the prospect exactly what working with you looks like before they spend a dollar.

The emails that ask for time get ignored. The emails that save time get replies.

2. Write for One Person Before You Scale

Every cold emailer we work with wants to talk about scale on day one. How many leads can we reach? How fast can we ramp? Can we send 50,000 emails in the first month?

Scale is the wrong starting point. Relevance is the right one.

Here is what we do for every new campaign. We build a list of 20 people who are an absolute perfect fit. Not 2,000. Not 200. Twenty. Then we write every single message by hand. No templates. No variables. Just one human writing to another human about a specific problem.

This sounds slow, and it is. That is the point.

When you write 20 emails by hand, you start noticing patterns. You realize that 15 out of 20 prospects share the same pain point. You notice that a specific phrase gets replies while another falls flat. You discover that the job title you thought was your buyer is not the one who cares about your offer.

Those patterns become the foundation for scale. Once you know what resonates with real people, you can turn those patterns into Clay variables and scale them across thousands of prospects. The personalization feels manual because it was born from manual work.

The sequence we follow:

→ Build a micro-list of 20 perfect-fit prospects

→ Write each email by hand, referencing specific details about their company

→ Track which angles, phrases, and data points get replies

→ Extract the winning patterns

→ Build Clay enrichment workflows that pull those same data points for every prospect

→ Scale to 1,000+ with messaging that still feels 1:1

Skipping the manual phase means you are guessing at scale. And guessing at scale burns domains, wastes data credits, and produces reply rates under 1%.

You can find verified emails for your initial micro-list of perfect-fit prospects here, for free:

Email Finder Tool

3. Your Lead List Matters More Than Your Copy

This is the lesson that surprises people the most. We have run campaigns where mediocre copy outperformed beautifully written emails by 5x. The difference was the list.

Sending your message to the right person at the right time matters more than crafting the perfect sentence. When you reach someone who is actively struggling with a problem you solve, the message writes itself. You do not need clever hooks or creative subject lines. You need a relevant signal.

Here is an example. We ran campaigns for an HR SaaS company that helps reduce employee churn. Instead of targeting every VP of HR at mid-market companies, we got specific. We looked for two signals:

→ Poor Glassdoor reviews (3 stars or below)

→ Low average employee tenure (under 2 years based on LinkedIn data)

If both signals were present, we assumed the company had a churn problem. The email was straightforward: "Noticed [company] has some retention challenges based on public data. We helped [similar company] reduce churn by 35% in 6 months. Worth a conversation?"

No tricks. No clever personalization. Just the right message to the right person at the right time.

The reply rate was 3x higher than the generic "VP of HR at 200-500 employee companies" list running the same offer with better copy.

Building signal-based lists takes more upfront work. You need to combine data from multiple sources. Clay handles the orchestration, pulling from providers like Prospeo, Wiza, and FullEnrich for contact data, while PredictLeads and Common Room surface intent signals and hiring data. The investment pays for itself in reply quality.

If you want to find people at specific companies with verified contact data, you can do it for free here:

People Finder Tool

4. Deliverability Is the Foundation

Good deliverability does not guarantee success. Bad deliverability guarantees failure.

You can have the best list, the best copy, and the best offer. None of it matters if your emails land in spam. We learned this the hard way early on when we scaled send volume too fast and watched reply rates collapse overnight.

Here is what we do on every campaign:

Warm-up every inbox for 30 days minimum. No exceptions. We use Instantly's built-in warm-up feature, which sends and receives emails from a network of real inboxes to build sender reputation gradually.

Use secondary domains only. Never send cold email from your primary domain. If something goes wrong, you do not want your main company domain flagged. We typically set up 3-5 sending domains per client, each with multiple inboxes.

Set up SPF, DKIM, and DMARC on every domain. These are email authentication protocols that tell receiving servers your emails are legitimate. Missing any of these is an instant red flag for spam filters.

Keep volume under 30 emails per day per inbox. This is a hard cap. Pushing beyond this damages sender reputation and triggers spam filters. If you need more volume, add more inboxes instead of increasing send rates.

Turn off open rate tracking. Tracking pixels are a known spam trigger. The data they provide is unreliable anyway since Apple's Mail Privacy Protection inflates open rates artificially. We stopped tracking opens entirely and saw deliverability improve across the board.

Send emails you would want to receive. This is the most underrated deliverability tactic. Spam filters are increasingly sophisticated. If your email reads like marketing, looks like a template, or contains spammy language, filters catch it. Writing genuine, conversational emails that provide value is the best long-term deliverability strategy.

You can check whether your emails will pass spam filters before sending, for free:

Spam Checker Tool

5. Speed to Reply Determines Your Close Rate

Positive replies are not the finish line. They are the starting line.

We tracked this across hundreds of campaigns and the pattern is clear: leads who receive a response within 5 minutes of replying are significantly more likely to book a meeting than leads who wait 24 hours for a follow-up. The prospect is thinking about your offer right now. They have context. They are at their desk. Wait a day and they have moved on to 50 other things.

The challenge is that cold email campaigns generate replies at unpredictable times. A prospect in London replies at 3am your time. A prospect in Sydney replies during your lunch meeting. You cannot manually monitor inboxes around the clock.

This is where Instantly's AI reply agents change the game. They monitor every inbox in real time, classify incoming replies by intent (interested, not interested, out of office, wrong person), and draft personalized responses based on the conversation context.

You can run them two ways:

→ Full autopilot: the AI replies instantly without human review. Best for high-volume campaigns where speed matters more than perfect tone.

→ Human in the loop: the AI drafts a reply and flags it for approval. Best for high-value prospects where you want a human to review before sending.

We use the human-in-the-loop approach for most campaigns. The AI handles the speed component by drafting a response within seconds. Our team reviews and approves it, usually within minutes. The prospect gets a fast, relevant reply without sacrificing quality.

Before we implemented this, our average response time to positive replies was 4-6 hours. After, it dropped to under 15 minutes. Meeting booking rates from positive replies increased by over 30%.

6. Putting It All Together

These five lessons compound. Lead with value so people want to reply. Write for one person first so your messaging resonates. Build signal-based lists so you reach the right people. Maintain deliverability so your emails land in the inbox. Reply fast so interested leads do not go cold.

The stack behind all of this is straightforward. Instantly handles sending, warm-up, deliverability, and AI reply management. Clay handles enrichment, signals, and personalization at scale. Prospeo and FullEnrich fill gaps on contact data. Everything connects through a workflow that runs continuously.

Cold email is not a one-time tactic. It is a system. And like any system, the quality of each component determines the output of the whole thing. Get the fundamentals right, and the results follow.