Cold Email Infrastructure vs. The 97% Strategy

Two Videos Worth Your Time This Week

I watch a lot of cold email content. Most of it recycles the same surface-level advice: personalize your subject line, keep it short, follow up three times. Nothing that actually moves the needle for someone trying to build a real outbound system at scale.

This week two videos stood out. One is a nearly four-hour full course on cold email infrastructure from someone running a legitimate agency. The other is a tight 22-minute breakdown from a sender who has processed over 50 million emails and has the client logos to back it up.

What connects them is more interesting than the tactical differences. Both are making a case that most people are failing at cold email for structural reasons, not copy reasons. One focuses on the sending side. The other focuses on the targeting philosophy. Together, they cover the two biggest leverage points most operators miss.

Let me break each one down.

Video One: The Infrastructure-First Framework

This is a full course, and it earns that label. The four sections, infrastructure, targeting and data, scripting, and appointment setting, map exactly to the four places cold email systems break down. Miss any one of them and the whole thing falls apart. That framing is correct.

The infrastructure section is the strongest part of the video. The core principle: make everything look as human as possible. That sounds obvious until you look at how most people actually set up their sending. One domain, one inbox, blasting 500 emails a day. That is not how humans send email, and the spam filters know it.

The solution presented is horizontal scaling. Instead of sending 1,000 emails from one inbox, you buy multiple domains, set up multiple inboxes per domain, and send 25 to 35 emails per day per inbox. Do the math: 35 to 40 inboxes at 25 to 30 emails per day gets you to 1,000 daily sends without triggering rate limits or burning your domain reputation.

The domain naming strategy is smart and worth copying directly. If your company is XYZ Marketing, you set up variations like getxyzmarketing.com, tryxyzmarketing.com, and xyzmarketinghq.com. Then within those domains you use different first names. Joe at getxyzmarketing.com. Joseph at xyzmarketinghq.com. This diversifies the infrastructure and signals to inbox providers that these are real individual senders, not a single automated blast.

I have run this exact setup at scale. The 25 to 35 emails per inbox per day number is conservative and that is the right call. I have seen people push to 50 and 75. It works until it doesn't, and when it breaks it takes your whole list with it. Conservative daily limits protect the long-term asset.

The DNS setup section covers MX, DKIM, SPF, and DMARC records. This gets glossed over in most cold email content because it is unglamorous technical work. It is also the thing that quietly kills deliverability for people who skip it. If you are not sure whether your DNS records are correctly configured, stop sending email right now and check. This is foundational. For a solid look at what tech you actually need in your sending stack, see the cold email tech stack guide.

On lead sourcing, the video recommends Apollo or Listkit to build your initial list, then validating with a tool like Million Verifier or NeverBounce before sending. The validation step is one most beginners skip. Sending to unvalidated lists increases bounce rates, and high bounce rates tank domain reputation fast. If you want a cleaner option, ScraperCity's email validator handles this before anything goes into your sequence.

The 4-email sequence structure mentioned, where you divide your daily send volume by the number of follow-ups to calculate net new prospects needed, is mathematically clean and practically useful. Sending 1,000 emails per day on a 4-touch sequence means you need 250 net new contacts per day, or roughly 5,000 per month. That is a concrete planning number most people never calculate. They just start sending and run out of contacts two weeks in.

On the appointment setting section: the video makes the point that replies mean nothing unless you convert them to scheduled calls. This is something I have seen trip up a lot of agency owners. They optimize for reply rate and then lose the lead in the back-and-forth of booking. The fix is simple: when someone responds with interest, your next message is one sentence with a Calendly link or specific time options. Not a paragraph. Not more selling. One sentence.

The tools recommended here are solid. Instantly and Smartlead are both legitimate options for managing multi-inbox sending at this scale. If you are just starting, Instantly has a cleaner onboarding experience. If you are running a larger agency operation, Smartlead gives you more granular control over sending behavior.

What to skip: the video briefly mentions sending 1,000 emails per day as the baseline goal, framed as the minimum to see results. For someone starting out, 1,000 emails per day is not the right starting point. You do not have the copy validated, you do not know your ICP response patterns, and you have not stress-tested your infrastructure. Start at 100 to 200 emails per day, learn what converts, then scale the sending volume. Scaling bad copy faster just accelerates failure.

Overall this is one of the more complete infrastructure walkthroughs I have seen in a free YouTube video. The depth on domain setup and horizontal scaling alone is worth watching if you are building or rebuilding your outbound system from scratch.

Video Two: The 97% Philosophy

This one opens with a claim that reframes the entire way most people think about cold email targeting. The idea: at any given moment, only 3% of your total addressable market is actively looking to buy what you sell. Most cold email strategies are built entirely around finding and converting that 3%. Intent signals, job change triggers, funding alerts, technology stack filters, all of it is an attempt to identify that 3%.

The argument against this is sharp. Everyone in your space is doing the same thing. They are all using the same Apollo filters, the same ZoomInfo intent data, the same LinkedIn signals. You are fighting in a red ocean over the same small group of already-identified buyers. Even if intent-based targeting worked perfectly, your competition for that attention is as high as it has ever been.

The proposed alternative: stop trying to predict the 3%. Instead, reach your entire total addressable market consistently so that when someone moves into buying mode, your name is the first one they think of. The goal shifts from converting active buyers to owning the memory of inactive buyers before they become active.

This is a legitimate strategic shift and I want to be specific about why it works. Cold email, at its core, is a frequency game disguised as a conversion game. The campaigns I have seen generate real pipeline over time are not the ones with the highest one-time reply rates. They are the ones that stayed in front of the right people long enough that when the timing changed, the response came in six months later referencing an email from the initial outreach.

The infrastructure redundancy lesson is the most operational piece in the video. The framework is three separate infrastructure sets: odd, even, and burner. Odd runs the first half of the month. Even runs the second half. Each set goes on warm-up when not actively sending. The burner set is a completely different infrastructure type, ready to rotate in when the primary two sets fail.

The key insight: it is not a question of whether your infrastructure is going to burn. It is a question of when. Domains age out. Inboxes get flagged. Providers have outages. If you are running a single infrastructure setup and it goes down, you cannot send for two to four weeks while you rebuild. For an agency running client campaigns, that is a catastrophic gap. The odd/even/burner system keeps you operational through what would otherwise be full stops.

I have had this happen. A domain batch I had warmed up over eight weeks got flagged after a provider policy change. Had a backup set ready. Total downtime was under 24 hours. Without that backup, it would have been three weeks minimum. Build the redundancy before you need it.

The lesson on matching sending infrastructure to recipient infrastructure is specific and valuable. The claim is that sending from a Gmail account to Gmail recipients versus sending from a mismatched infrastructure type can produce a 3x to 16x improvement in reply rates. The mechanism is that inbox providers trust their own systems and deprioritize messages from external infrastructure they do not recognize as trusted senders.

The no-links-in-first-email rule is one I agree with completely. Links trigger phishing filters. This includes tracking pixels for open rates. If you are tracking opens via a pixel, you are inserting a link into every email. That link flags security filters even when your domain is clean. The recommended format instead of a hyperlink is to write out the URL with formatting like fixer (dot) com or rb2b.com spelled out without making it clickable. The goal of the first email is to start a conversation, not send someone to a landing page. If they are interested, they will Google you anyway.

The data source lesson is where the video gets genuinely contrarian. The argument: everyone is using the same Apollo lists with the same filters, which means the same contacts are getting hit by every competitor in a given space simultaneously. The people on Apollo know they are on Apollo. Their inboxes reflect it. The edge is in proprietary lead lists that no one else has access to.

This connects directly to something I have built around at ScraperCity. When you are working from the same publicly available B2B database as every other sender in your niche, your reply rate ceiling is set by the noise floor, not by your copy. Building lists from sources your competitors are not using is a structural advantage that compounds over time. The ScraperCity B2B database covers a substantial portion of the US B2B contact universe, which matters when the standard tools are oversaturated. You can also build tighter, more targeted lists using the people finder tool for specific segments your competitors are not targeting.

The copy lesson is the one that will frustrate people who want a formula. The conclusion from 50 million emails sent is that there is no universal copy framework. Short punchy two-liners crushed it for a tech-forward SaaS audience. The exact same approach underperformed for an HR tech platform where long-form storytelling copy was the winner. The only way to know what works is to test multiple genuinely different styles simultaneously, not small tweaks to the same template. The video mentions launching 10 to 15 copy variants at once for certain clients. That is not a typo. If you want frameworks worth testing, the top 5 cold email scripts break down the structures that tend to perform across different buyer types.

What to challenge: the claim that intent signals have never shown up in data as effective is stated confidently but without specifics. In my experience, intent data works in narrow contexts. If you are selling something with a short buying cycle and a clear trigger event, like a tool that activates when a company makes a specific hiring decision or technology change, intent-based targeting can meaningfully increase conversion from that 3% window. The broader point about not exclusively targeting the 3% is correct. But dismissing intent signals entirely as ineffective is probably too strong a claim without more context on which intent signals and which product categories.

The Pattern Both Videos Are Making

Here is what connects these two videos. Neither one is primarily about copy. Both are making structural arguments about why most cold email systems fail before a single word of the email matters.

The first video says: your infrastructure determines whether your emails land in the inbox at all. Get the domain setup, the horizontal scaling, and the warm-up right before worrying about subject lines.

The second video says: your data source and your targeting philosophy determine whether the right people see your emails at all. Build lists your competitors do not have, and stay in front of your full addressable market rather than hunting the same 3% as everyone else.

Together they are making the same argument from different angles. Infrastructure and targeting are multipliers. Copy is an input variable within a system. Most people spend 80% of their time on copy and almost none on the underlying system. That is why their results plateau.

The practical implication: audit your infrastructure before testing new copy. Check your domain health, your DNS records, your per-inbox send volume, and your backup infrastructure status. Then look at where your list is coming from. If every lead in your system came from the same Apollo export as your three closest competitors, that is a targeting problem no amount of copy optimization will fix. Tools like the Apollo scraper let you get more targeted with how you build and segment those lists rather than working from generic exports.

For the actual follow-up mechanics once replies come in, the cold email follow-up guide covers the sequencing and response handling that converts interested replies into booked calls.

What to Implement This Week

One thing from each video, implementable in under a day.

From the infrastructure video: calculate whether your current per-inbox daily send volume is within the safe range. If you are sending more than 40 emails per day from any single inbox, dial it back to 25 to 30 and spread the remaining volume across additional inboxes. This one adjustment has recovered deliverability for more outbound operations than any copy change I have seen.

From the 97% video: add a plain-text email to your current sequence, zero links including the unsubscribe footer if your tool allows it, zero tracking pixels, URL spelled out without a hyperlink. Send that version to a segment of your list and compare reply rates over two weeks. The deliverability difference is measurable and the change takes about 15 minutes to set up.

Both of these are infrastructure-level fixes. Neither requires you to rewrite a single line of copy. That is the point.