Volume vs Personalization: Cold Email Breakdown

I Watched Two Cold Email Experiments So You Don't Have To

This week two videos caught my attention. Not because they were polished. Not because the channels were massive. But because both creators actually ran real campaigns and shared real numbers. That's rare. Most cold email content is theory dressed up as experience. These two guys went out and did the thing.

What makes this interesting is that they took completely opposite approaches. One sent 10,000 emails per day for 30 days. The other sent 6,474 total emails over several weeks, each one personalized at a level most people don't bother with. Same goal. Wildly different execution. The results tell you something worth understanding.

Let me break both down.

Video 1: Aaron Shepherd Sends 10,000 Cold Emails Per Day for 30 Days

Channel: Aaron Shepherd | GrowthFlare (1,720 subscribers)
Views: 3,103

What He Actually Did

Aaron set out to send 10,000 cold emails per day for a full month. That's roughly 300,000 emails total, or about 210,000 sends per month after accounting for weekends and warm-up periods. To do this without getting everything burned, he built out a serious infrastructure: 114 domains, a 70/30 split between Outlook and Google inboxes, two inboxes per domain, and a two-week warm-up period before going live.

He used .info domains at around $3 each instead of .com domains at $10, specifically because at that volume you rotate through domains fast enough that the cost difference matters more than the slight reputation difference. His total domain cost came to roughly $367 as a one-time fee. Monthly inbox costs ran about $600 for Google accounts and $400 for Outlook accounts through a reseller called Premium Inboxes.

For leads, he targeted B2B lending firms through Apollo, using ChatGPT to generate keyword lists and job title filters, then mapping those directly into Apollo search criteria. He acknowledged he'd need around 105,000 verified prospects per month to sustain the send volume, which meant testing multiple segments throughout the 30 days.

His sending platform was Email Bison, and he configured everything to autoconfigure DNS, SPF, and DMARC settings automatically through the inbox provider.

Early in the video, he mentioned getting around 10 interested responses in the first batch, which he called a solid meeting book rate. The exact percentage was cut off in the transcript, but the framing suggests it was somewhere around 0.1% of sends converting to interested replies.

What I Agree With

The infrastructure thinking here is correct. If you're going to send at volume, you need to distribute across domains and providers. Putting everything through one domain at this scale is how you end up with your whole sending reputation destroyed in 48 hours. The 70/30 Outlook to Google split makes sense from a cost standpoint, and two inboxes per domain is a reasonable conservative setup.

The .info vs .com decision is also defensible. Aaron's framing that the cost savings outweigh the minor deliverability difference is something I've seen play out in practice. When you're rotating domains frequently, you're not building long-term domain reputation anyway. You're burning through infrastructure as a cost of doing business at scale.

Using Apollo filtered by job title, company keywords, and verified email status is the right starting point for B2B prospecting. If you want to go deeper on building lead lists like this, our Apollo scraper at ScraperCity can help you get more out of that process.

What I'd Push Back On

Here's where I have a real concern with the entire framing of this experiment. Sending 10,000 emails per day is not a strategy. It's a volume play that assumes your offer and copy are good enough to get results at scale without optimization. Aaron admits early in the video that a previous 50,000-email week didn't produce the results he wanted. His response was to send more. That's not necessarily the right lesson to take from underperforming results.

In my experience, when a campaign isn't converting, the answer is almost never more volume. It's better targeting, a sharper offer, or tighter copy. I've helped over 14,000 entrepreneurs generate sales meetings, and the ones who try to brute-force their way to results through volume almost always hit a wall. The ones who obsess over the offer and the segment tend to compound results over time.

I've personally sent millions of cold emails. At no point did 10,000 per day feel like the answer to low reply rates. What mattered was finding the segment where the offer clicked and the pain point was real. That's something you can test with 200 emails. You don't need 10,000 a day to figure it out.

Also, the economics here deserve scrutiny. You're spending roughly $1,000 per month on infrastructure before you write a single email. That's not outrageous for an agency running multiple clients, but for a solo operator or early-stage business, that budget committed to a well-targeted list of 500 people with a strong offer will outperform the 10,000-per-day approach almost every time.

What's Worth Implementing

The infrastructure setup Aaron describes is genuinely useful if you're already past the point of testing and you know your offer converts. The domain purchasing process on Pork Bun, the 70/30 inbox split, the two-week warm-up before launch, the autoconfigured DNS settings through a provider like Premium Inboxes. That's a solid operational playbook for scaling a proven campaign.

The ChatGPT-assisted Apollo keyword generation is also worth stealing. Using a prompt to generate job titles, company keywords, and size filters that you can paste directly into Apollo search criteria saves time and often surfaces targeting angles you wouldn't have thought of manually.

But start with 200 emails to test the offer. Then scale. Don't scale first and optimize later. That's backwards.

If you want the email scripts that actually drive replies before you start thinking about volume, check out my top 5 cold email scripts here.

Video 2: 30 Minutes to President's Club Sends 6,474 Personalized Emails

Channel: 30 Minutes to President's Club (16,500 subscribers)
Views: 1,313

What They Actually Did

This one is a different animal entirely. Armand Farrokh from 30 Minutes to President's Club partnered with Everett from Clay to run what they called the biggest personalized email campaign they'd ever launched. Their list started at 224,000 contacts: Armand's LinkedIn followers, Nick's LinkedIn followers, and their mailing list combined. From that, they used Clay to filter down to 2,894 US-based sales leaders with at least 10 reps on their team. Then they filtered further to only target sales leaders who had at least three team members that had already consumed 30 MPC content, which brought them to 2,158 people.

They connected HubSpot to Clay to identify which contacts on each domain had converted on 30 MPC content, pulled the names of up to three of those reps per contact, and used that data to personalize the subject line and email body. The subject line format was the names of the three reps: something like "Matt, Pranav, and Coral." The email body followed a four-part structure: personalization trigger, problem tied to the content the leader consumed, one-sentence solution, and a CTA asking if they'd be open to seeing a training program example.

They set up 20 sending domains, warmed them for two weeks, and launched with conservative daily send limits of 25 to 50 emails per domain. They sent 5,224 emails total to 195 leads across a three-week window. They got 57 replies for a 1.1% reply rate. Of those, 13 were positive, which is 22% of replies converting to interest.

What I Agree With

The targeting logic here is exceptional. They didn't just filter by job title and company size. They filtered by behavioral signals: people who had already consumed content, whose team members had also consumed content, who were in the right role, at the right company size, in the right geography. That's not cold outreach in the traditional sense. That's warm targeting with cold email mechanics layered on top. The prospect already has a relationship with the brand. The email is acknowledging that relationship.

The subject line using rep names is clever and worth studying. It's not a trick. It's a relevance signal. If I'm a VP of Sales and I see the names of three of my reps in the subject line of an email, I'm opening that email. Not because it's gimmicky but because it's immediately relevant to something I care about: my team.

The email structure they used maps closely to what I teach in Clay-powered outreach. Lead with the trigger, connect it to a problem, solve it in one sentence, give them an easy CTA. That's solid cold email architecture. If you want the underlying framework for follow-ups after that first touch, my cold email follow-up templates walk through exactly how to sequence after the first reply.

The three-email sequence structure is also right. Email one does the heavy personalized lifting. Email two is a one-sentence bump. Email three is a breakup that asks for a thumbs up or down and offers to check back in 90 days. That's a complete sequence that respects the prospect's time without abandoning the thread prematurely.

What I'd Push Back On

The results were underwhelming relative to the setup, and I think Armand knows it. He addressed it directly in the video, saying that a warm 30 MPC audience should have converted closer to 40 to 50% of replies as positive rather than the 22% they saw. He's right. When your list is this warm, when three of their direct reports have already consumed your content, and when you're the brand teaching them how to sell, a 1.1% reply rate is soft.

In my experience, the reason this happens is that personalization at scale through AI still has a ceiling on quality. The Clay-generated emails were personalized with rep names and content references, but if the actual email body reads like it was written by a formula rather than a person who actually did the research, sophisticated sales leaders notice. They've seen enough AI-generated outreach at this point to recognize the pattern.

Another issue is the send volume per domain. Sending 25 to 50 emails per day from two to three week old domains is conservative, but they still sent 5,224 emails to 195 leads. That math suggests they were sending multiple touches to the same people across a sequence. That's correct. But with 2-week-old domains and that kind of saturation, deliverability issues compound. Some of those 57 replies may represent only the portion of sends that actually landed in primary inboxes.

The comparison to ads that Armand implicitly makes is worth questioning too. They invested significant Clay credits, HubSpot integration work, and Everett's time to personalize at this level. For 13 positive replies. The ROI math only works if those 13 convert to deals at a high rate given the premium course offering. If your product is lower ticket, this level of personalization infrastructure is likely overkill.

What's Worth Implementing

The filtering logic is the most transferable thing from this video. Start with a big list of people who have some warm signal, a LinkedIn follower, a content consumer, someone who attended a webinar, someone who downloaded a lead magnet. Filter that list by ICP criteria: job title level, team size, geography. Then filter again by engagement signals: who has three or more team members who've consumed your content, who opened your last three emails, who clicked something. Now you have a list that's 2,000 people instead of 224,000, but every person on it has demonstrated relevance. That's the list worth personalizing.

Using Clay to connect your CRM data to your outbound prospecting table is genuinely powerful. The ability to look up associated contacts by domain and pull conversion events from HubSpot into a personalization variable is not something most people are doing. It's not simple to set up, but if you have an existing content audience or customer base, it unlocks a level of relevance that pure cold outreach can't match.

The subject line format of using names directly is something worth testing regardless of whether you use Clay. If you know three people at a company who are relevant to the email, put their names in the subject. It's not manipulation. It's proof that you did your homework.

The Pattern Across Both Videos

Here's what jumps out when you look at these two experiments side by side: both creators ran real campaigns, shared real numbers, and both got results that were solid but not great. Aaron's volume play produced meetings but didn't hit the rate he was hoping for. Armand's personalization play produced a 1.1% reply rate from an audience that should have been significantly warmer than a cold list.

The lesson isn't that volume is wrong or that personalization is wrong. The lesson is that the offer and the segment are still the highest leverage variables in any cold email campaign. Neither volume nor personalization can fully compensate for reaching the wrong person with the wrong message at the wrong time.

What Aaron should have done before scaling to 10,000 per day is test his offer with 200 sends across three different segments and identify which one converted at 2% or above. Then scale. What Armand's team did well was behavioral segmentation but the actual email copy may have been diluted by the AI personalization layer, which flattened the human quality of the message.

I've seen this play out in consulting sessions consistently. One client sent 667 cold emails with a 50% open rate and 74 replies. Of those 74, the issue wasn't volume and it wasn't personalization. It was that the offer framing in the response sequence was too vague. When we tightened the follow-up reply, meetings started booking. The infrastructure was fine. The copy was the bottleneck.

If you want to see the tech stack I actually recommend for cold email right now, including what tools work together and what's worth paying for, check out the cold email tech stack breakdown here.

What to Actually Do This Week

Don't start by building 114 domains. Don't start by connecting your CRM to Clay and building a behavioral segmentation waterfall. Both of those are second or third stage moves.

Start here: pick one segment of 200 people where you have genuine reason to believe the offer is relevant. Write one email. Send it from one domain with one inbox. Measure the reply rate. If it's above 2%, you have a working campaign. Now you can scale the infrastructure. If it's below 2%, change the offer or change the segment before spending another dollar on domains or personalization tooling.

The infrastructure Aaron built is real and functional. The personalization system Armand used with Clay is sophisticated and genuinely impressive. But both are solutions to a problem you can only identify after the offer is proven.

Get the offer right first. Then decide whether you need volume, personalization, or both to scale it.

If you want the email copy frameworks to build that first working campaign, the killer cold email templates are a solid starting point. These are the structures I've used across millions of sends and that I've coached over 14,000 entrepreneurs to use when building their first working cold email campaign.