The Cold Email Playbook Smartlead used to hit $20M ARR

Cold email isn’t dead—but it’s been fundamentally transformed. What used to be a game of volume and brute-force tactics is now dictated by intelligent filters, content relevance, infrastructure hygiene, and increasingly, AI models judging your message before a human ever sees it.

In a deep-dive interview, Vaibhav Namburi (known as “V”), the founder and CEO of Smartlead, shared a granular, data-backed guide on how to run cold outreach that actually works in 2025.

Having bootstrapped Smartlead into a $20M+ ARR business with minimal headcount, V isn’t just another tool builder—he’s helping orchestrate hundreds of thousands of cold email campaigns at scale. This post is your complete guide to what he’s learned. If you do outbound, this is your new playbook.

The Email Tool That Cracked the Primary Inbox Algorithm

Smartlead's mission is simple: help teams consistently reach their leads’ primary inboxes.

Whether you're sending 10 emails or 10,000, if they land in spam or promotions, they won't get read. V and his team built Smartlead to address exactly that problem—with infrastructure, testing tools, and a whole lot of insight into how email systems actually work today.

Smartlead started with outbound as the core growth engine. In the company’s first year, outbound emails generated the majority of their revenue.

As Smartlead matured, outbound became part of a broader strategy that included paid advertising, social media, SEO, content marketing, and branded events.

But V emphasized that outbound still plays a crucial role. It now serves as a multiplier—accelerating reach, supplementing brand, and opening doors that wouldn’t otherwise open.

Infrastructure First: The #1 Thing to Nail Before Hitting “Send”

The first non-negotiable element of successful cold email is infrastructure. If your technical setup is wrong, no copy hack will save you.

Previously, teams could send 40–50 emails per day from a single mailbox. That era is over. Today, the norm is 20 to 27 emails per day per mailbox, especially for domains that are still warming up. Some teams send even fewer—just 2–5 emails daily—when deliverability is paramount.

Smartlead sees many clients leveraging shared Outlook mailboxes or building custom configurations with Microsoft Azure to manage scale without increasing risk. These setups help teams mimic human behavior while maintaining sending volume. Importantly, they also offer a way to distribute risk.

This is where the concept of tenants becomes important. In Microsoft 365, a “tenant” refers to your entire workspace. If one domain or mailbox in your tenant gets flagged, your entire tenant’s reputation is affected. That means every other mailbox under that tenant is also penalized. Smartlead encourages outbound teams to isolate their infrastructure—especially when targeting high-risk or enterprise accounts. Having dedicated tenants for cold outreach can prevent collateral damage to your core domain and help you recover more easily if something goes wrong.

And when you're targeting the big leagues—think Nike, Amazon, or Apple—V warns that enterprise-grade filters like Barracuda and Proofpoint don’t just filter; they punish. Once you’re flagged, you’re not just flagged for that recipient—you’re blacklisted across the entire organization’s mail server. The fix? Set up entirely separate domains and tenants for enterprise outreach. Treat them like a separate product line with separate risk thresholds.

Spam Filtering Has Entered Its Third Evolution

Then there is the evolution of spam filtering. Originally, email providers focused on whether your domain passed SPF, DKIM, and MX checks. Then they evolved to look at behaviors like bounce rates, reply rates, and domain volume.

Now we’re in phase three: content-based filtering. Email services have begun analyzing message content itself—sentence structure, phrasing, and historical signals on whether that content has previously been flagged.

If a message (or even part of a message) was previously marked as spam, future emails with similar phrasing may go to spam by default.

Real-World Proof: Spam Propagation in Real Time

To test this, Smartlead ran an experiment. They sent the same message to 10 internal inboxes. When the first email was marked as spam, refreshing the remaining inboxes showed 3–4 of the other emails disappearing into spam immediately.

This wasn’t caused by domain problems, HTML formatting, or blacklists—it was caused purely by content repetition.

Spam filters today are driven by a blend of security layers, user behavior, and AI-trained language patterns. That means your copy needs to be treated like code: it must be dynamic, varied, and safe by design.

And here’s something many overlook: open tracking pixels themselves—commonly used to see if someone “opened” your email—can trigger spam filters too. Since these pixels rely on HTML and are linked to static behavior, they increase the fingerprinting risk. V strongly recommends not using open tracking as a performance metric anymore. If you need insights, inbox placement testing is the far safer and more accurate route.

Even Signatures Can Get You Flagged: The Tiny Detail that Sinks 1,000s of Cold Campaigns

One surprising revelation: static content in your signature block can damage your deliverability.

V shared a story about a Smartlead customer who saw a meaningful improvement in inbox placement after removing the phone number from their email signature. The reason was that the phone number had been included in thousands of other messages—many of which were ignored or flagged.

Static details like phone numbers, booking links, or logos may appear safe, but their repeated use trains filters to associate them with low-quality email. To mitigate this, vary your closing statements and test signature variants as aggressively as you test subject lines.

And if you're using links (even in your signature), avoid using the same static URL over and over across your campaigns. V called this link staticization—and it’s a hidden landmine. Even if it’s just https://calendly.com/you repeated use across your sender accounts will degrade performance. When you do include links, vary them dynamically or hold them back for follow-ups after inbox trust has been built.

Tools to Monitor and Improve Inbox Placement

V recommends using both Microsoft Defender and Google Postmasters to track your sender health and inbox performance.

Microsoft Defender (available in premium tiers) allows you to see whether your email landed in the recipient’s spam, junk, or primary inbox—especially helpful for Outlook-based leads.

Google Postmasters provides data on:

• Spam complaints

• Authentication failures

• Delivery errors

• IP and domain reputation

A practical tip: export all the messages that got flagged as spam, dump them into ChatGPT, and ask:

“What are the most common word patterns or phrases across these?”

This helps you identify spam-triggering language based on real-world outcomes—not theoretical lists.

Understanding Spam Complaint Thresholds: Avoid the 0.3% Death Zone for Cold Email

A year ago, Google introduced a policy limiting spam complaints to 3 per 1,000 emails—specifically for Gmail users.

If your domain crossed this line, it risked getting downgraded or blocked entirely.

While the rule was later walked back for B2B outreach, the principle still holds: keep your spam complaint rate under 0.3% to stay in the clear.

Personalization at Scale: Spintax and Smart Content Generation

In a world where content repetition gets you penalized, Spintax is not just a nice-to-have—it’s essential.

Smartlead has long supported nested spintax, which allows users to dynamically vary parts of an email. Not just greetings like “Hi” vs. “Hey,” but entire sentence structures, qualifying statements, and CTAs.

Good vs. Weak Spin Tax

Weak:

“Hi | Hey | Hello {first_name}”

Strong:

“Good morning {first_name}, I came across your recent work at {company} and had a few thoughts.”

Even if the core message is the same, unique phrasing improves your chance of inboxing. Smartlead now includes an AI-based spintax generator: paste your core copy, hit a button, and get dozens of variants built for deliverability.

And here’s the critical distinction: spintax is not A/B testing. Spintax is for beating spam filters through randomization. A/B testing, on the other hand, is about learning which copy actually converts. Don’t confuse the two—they solve very different problems.

Subject Line Strategy: Don’t Randomize—Test Methodically

Spintax doesn’t work well for subject lines, because you can’t track performance easily. Instead, V recommends A/B testing up to 90 subject line variants.

Subject line best practices in 2025:

• Use all lowercase

• Keep it under 4–5 words

• Make it feel like an internal message

Examples:

• “quick idea for you”

• “just saw this”

• “worth a look?”

The goal is to mimic internal communication, not marketing blasts.

Messaging Strategy: Relevance > Education

Most cold email mistakes happen in the first two sentences. Sales reps often try to educate the recipient—but if you’re emailing a CMO, don’t explain how marketing fraud works. They already know.

V calls this the presumed knowledge curse. When you make assumptions about what the reader does or doesn’t know, you lose them.

Instead:

• Lead with insights specific to their company

• Reference pain points they already feel

• Use copy that reads like it's written by a peer, not a pitchbot

Avoid spammy terms like:

• “hijack”

• “fraudsters”

• “alert”

• “urgent”

If you must use them, soften them with context or substitute with language that’s less likely to trigger filters.

Also, avoid writing emails with a one-size-fits-all lens. One of the most fascinating concepts V introduced is that mailboxes have personas. A message about hair products might land in spam for him, but go straight to primary for someone else who engages with similar topics. Relevance isn’t universal—your copy must match the recipient’s behavior profile.

The Sequence Myth: Why Most Campaigns Die After Two Emails

Forget the five-touch sequence you read about in 2018. Today, most inboxes will tolerate two cold emails—and that’s it.

Smartlead’s customers are increasingly adopting a 1:3 or 1:4 lead-to-sequence ratio. That means one lead receives three or four emails max. Beyond that, performance drops and risk increases.

Here’s why: even if your reply rate is 20%, you’re still sending 80% of your messages into the void. Do that for six months straight and your domain reputation will crash.

Fewer emails, more precision, and better targeting wins in 2025.

When to Send the Loom—and When Not To

Links are risky in cold outreach, especially in the first email.

Never include:

• Static URLs (e.g., “visit our website”)

• Embedded links

• Images or logos

• Phone numbers

These static assets get flagged fast. Once you build some inbox trust, use tools like:

• Loom

• SendSpark

• Fixer

These platforms allow for personalized video messages with dynamic links—which are less likely to be caught by spam filters. Smartlead recommends saving these for the second email onward.

Stop Obsessing Over Open Rates

Open tracking is nearly obsolete thanks to Apple Mail Privacy Protection, Gmail restrictions, and growing user control. Don’t use open rate as your north star.

Instead, use inbox placement testing. Smartlead offers test campaigns that send your message to hundreds of controlled mailboxes to determine where your email actually lands: primary, promotions, or spam.

Smartlead is developing a next-gen inbox testing feature. It will simulate deliverability across recipient personas:

• 500 inboxes modeled after CEOs

• 500 as sales leaders

• 500 as engineers or marketers

Why? Because spam filters don’t just assess content quality—they assess relevance to recipient behavior.

An email about beauty products might land in promotions for one user, but go to primary for another. Persona modeling will reflect this nuance.

Validating Leads and Filtering the Right Targets

Good cold email starts with good targeting. Even the best copy and infrastructure won’t save you if you’re emailing the wrong person.

Here’s the ideal tech stack V sees used by top-performing outbound teams:

• Apollo: For raw lead sourcing and enrichment

• Clay: For ICP filtering, AI personalization, and dynamic copy creation

• Lead Magic: For email validation (Smartlead also has a built-in validator, but V praises Lead Magic’s accuracy)

Together, this stack helps filter out bad leads, minimize bounce rates, and send the right message to the right people.

The 3-step playbook for cold email in 2025:

• Beat the AI, then the human. Write emails that avoid spam filters first and appeal to people second.

• Avoid being spam by not acting like spam. That means low volume, high personalization, and constant variation.

• Don’t fix what’s working.

The playbook for 2025 isn’t about gaming the system. It’s about sending better messages, built on solid infrastructure, to the right people—with the right context.

Apply these principles, and you won’t just reach inboxes. You’ll start more conversations that actually go somewhere.

Top Tweets of the day

1/

Evals are the gold standard for judging an AI product.

If you want to know if your prompt is good or not, you have to test it on evals. The better the evals, the more you know if your prompt worked or not.

This is similar to testing in software development world. Testing gives confidence on your software.

2/

Rork went from $0 to $1M ARR in 3 months with 4 people. This is one of the reasons why. You can re-build a single-utility mobile app in <1 week.

My favorite tech meme/quote is by @iamdevloper "The dirty secret: We’re all just building CRUD apps." where CRUD stands for Create, Read, Update, Delete.

All apps do basically CRUD on a higher-level. Take BetterPic AI for example. It creates an image from a prompt, reads the image to display it to the user, updates the image based on the prompt, or simply deletes the images. And CRUD applies to settings, payments, etc... too.

3/

TikTok is where B2B purchases will be made in future. B2B and B2C will probably have a massive overlap as humans love to waste time scrolling when they are bored.

Rabbit Holes

1. Taking a site from 0 to 10.9M in 120 Days with AI SEO + GEO by Vasco

2. How Grey Goose Used Psychology to Become a $4B Brand by Choice Hacking

3. Growing a scraping API to $10k+MRR in 12 months by James Fleischmann

What’d ya think of today’s newsletter? Hit ‘reply’ and let me know.

Do me a favor and share it in your company's Slack #marketing channel.

First time? Subscribe.

Follow me on X.

More Startup Spells 🪄

1. Kilo Code Playbook: Beating Cursor, Windsurf, Claude Code with Guerrilla Marketing Tactics (LINK)

2. Riverside FM's Dwell Time Using Free Transcription Service (LINK)

3. Copywriting Hack So Good That It Feels Illegal (LINK)

4. Vista's Billion-Dollar Private Equity Strategy for Customer Selection (LINK)