Learn / Results
Three cold email campaign teardowns, with the real numbers
Most of what gets posted as a cold email case study is a calendar screenshot and a number with no timeframe on it. So here are three of ours taken apart: the offer, the list, the copy, the qualification, and the funnel each one produced. 42 calls in 30 days for an e-commerce exit consultant. 21 calls in 21 days for a funding enablement business. 47 calls in 45 days for an M&A advisory firm.
Three offers, three lists, one shape underneath. The video is the first teardown end to end. The other two are written out under it.
Teardown one: an e-commerce exit consultant, 42 calls in 30 days
The infrastructure. Their previous agency sent everything from one domain, which does to deliverability what you would expect. We rebuilt at roughly 125 sending accounts split evenly between Google and Outlook: 40 Google accounts across 20 lookalike domains, two per domain, all forwarding to the primary, plus 85 Outlook accounts on one domain. Every account warmed two weeks minimum before sending, at 1.5 times its sending limit, and stayed warming for the life of the campaign. Google publishes what it expects from bulk senders, and warming is the cheapest way to meet it.
The list. This started far more complicated than it ended. The first build validated each company as a real e-commerce store using signals like a shipping policy and an affiliate program, then estimated revenue from site traffic and average order value, because revenue data inside the big databases is not accurate. It came back too small, and too much of the estimate was wrong or missing. So it got cut to ONE question: is this an e-commerce store. The monthly target then moved from 10,000 prospects to 30,000.
The offer.What they had was a description of their service, which is not an offer. A stranger needs to know exactly what they get, in what timeframe, with how little resistance. So we built a free exit consultation: how prepared the owner was to sell, a valuation estimate, and industry-specific exit data and multiples. Behind it sat a softer option for anyone who would not give up the time, a seller's planning guide with follow-up sequences on the other side of it. Building one for your own service is the front-end offer playbook.
The copy. Six lines, longer than we normally write, each with a job. A curiosity question about where they stood on exiting, left as an open loop so they had to read on. Social proof: who we had helped exit and how long it took. An elaboration line spelling out what the consultation included, so the value was obvious before they said yes. The ask. Then a second CTA, which is where this campaign got its lift: even if you are not ready to sell, here is why the conversation is worth having. Then a PS turning the opt-out line into a joke about not being ready to sip margaritas on a beach yet. It took the sales pressure out and made it obvious a person wrote the email.
The qualification. At 1,000 emails a day the problem stops being interest and becomes sorting it. Every interested reply hit a pre-call form: would you consider selling in the next 12 months, approximate revenue, percentage of revenue from e-commerce, approximate profit, inventory value, current debt. Anyone who did not clear went into nurture instead of onto the calendar, which is a pre-call qualification job of its own.
What it returned. Over 30 days: 565 replies, 79 interested, just under 15% of everyone who wrote back, 42 calls booked, 2 closed. The planning numbers going in were a 2% reply rate, 10 to 20% of replies interested, half the interested booked. What to expect on your own offer is in the reply rate benchmarks.
Teardown two: a funding enablement business, 21 calls in 21 days
The offer. Funding support in the innovation space, which reads as saturated the moment you call it grant support. So we sold the outcome instead of the service, priced as a percentage of the funding actually won. Performance-based is hard to say no to, and here the offer did enough of the work that everything downstream got easier.
The list. Decision makers only: founders, CEOs, presidents. Then two angles into the market instead of one. Geography first, because California is genuinely a hub for innovation work, so that campaign was built on region rather than industry. Then industries innovative by nature, biotech among them. Then a floor of 50+ headcount, because the offer carries an upfront cost. Apollo into Clay, a waterfall through other providers wherever Apollo had no email, verified through MillionVerifier, company names normalized. 4,600 prospects from that list, around 7,000 across all the campaigns.
The copy. Subject lines two or three words, lowercase, shaped like a question and nothing like marketing material: {{first_name}} grant?, grant idea, innovation grant. The opening line led with relevance rather than a generic AI compliment, and we ran two versions: a broad line about the work they were doing, and an AI-written line specific to their project. When the offer is this relevant, the specific version stops earning its cost.
The follow-up. Everyone got two steps, and only the people who showed interest moved into the four after that. Sending five emails to someone who does not want the thing is how you get marked as spam and lose the inbox for the people who did.
The part that doubled the bookings. Every interested reply got enriched for a mobile number through a provider waterfall, formatted into a click-to-call link, and pushed to our SDR in Slack so the prospect got a call inside 5 minutes. Booking half your interested replies is the rule of thumb on email alone. With calling attached, this ran closer to 8 in 10.
What it returned. Roughly 12,000 emails over 21 days to just under 7,000 net new prospects, a little over a 2% response rate, 29 interested replies, 21 calls on the calendar.
Teardown three: an M&A advisory firm, 47 calls in 45 days
The problem with this one.Replies are easy when you are offering to buy somebody's business. Done carelessly, most of them come from owners of terrible companies, which is precisely why they are willing to sell. So the campaign was built backwards on purpose: wide at the top, hard filtering after the reply.
The list. Apollo, founder or owner, US only, 51 to 100 employees, and the filter most people skip: minimum 10 years in business, because a company that has survived a decade is established enough to be worth buying. Industry and keyword left deliberately blank. Cut to first name, last name, email, organization and website, then every address verified before anything else, so no AI credits got spent on prospects who cannot receive mail. Only then did Claygent, running a cheap model, read each website and classify its industry.
The offer.You cannot ask an owner outright whether they want to sell. Too direct, and most people say no even when they are interested. So we used what I call strategic bait: implying slightly more intent than exists. Here, we already have a buyer interested in businesses like theirs. It is true, because the client is an M&A firm and that is the job, and it flips the frame into urgency, curiosity, and an owner who feels chosen rather than prospected.
The copy. Subject line two or three words, vague and curious, something like {{first_name}} thoughts. The body is one first line plus a two-part CTA:
We have a buyer looking for businesses in the {{industry}}industry, given their current multiples. Since you have been running yours for over 10 years, would it be worth a quick intro? Even if it's just to see what the offer on the table could be.
The industry token is the Claygent output, which is what lets one list cover many industries and still read like it was written for one. The multiples line implies the opportunity is real, so use it carefully, because plenty of industries do not have flattering multiples. And the second half of the CTA stacks interest: if the intro does not catch them, curiosity about the number might.
The qualification. Three questions on a pre-call form: their financials, when they are actually looking to exit, and an optional one on why. Friction is a lever, not a setting. Tighten it too far and you cut the good calls out with the bad, so use the least friction that still tells you whether a call is worth taking.
What it returned.Over 45 days: just under 40,000 emails to 20,000 prospects at two emails each, 789 replies at just under 4%, 164 interested at almost 21% of those, 47 calls booked. Several have since moved into the next stage of the client's process.
Three offers, three lists, one habit: send wide, qualify late.
What repeats in all three
Take the verticals out and the same 4 decisions carry every one of them.
- The list goes wide, the filter goes late. One build spent weeks scoring revenue up front and got deleted. Another left the industry field blank on purpose. Cold email is cheap reach, so use it as reach and qualify after somebody raises a hand.
- The offer is what they reply to. An exit consultation with real multiples. A fee paid out of funding won. A buyer who already exists. None of these won on a clever sentence.
- Follow-up is earned, not scheduled. Two steps for everyone, the rest only for people who showed interest. Deliverability decision as much as a conversion one.
- Something owns the gap between interested and booked. A pre-call form on two of them, a phone call inside 5 minutes on the third. That gap is where most campaigns quietly lose half their calls and blame the leads.
One caveat before you copy any of it: the offer sets the reply rate, so none of these are targets to hold yourself to. That campaign closed 2 deals off 42 calls in a month, and in M&A the economics of 2 closes are fine. In your business they might not be. Judge on booked calls worth taking, never on how good the reply rate felt.
tl;dr: three offers, three lists, one shape. Send wider than feels comfortable, make the offer worth answering, and treat everything between the reply and the booked call as its own job.
FAQ
What should a cold email case study actually show you?
Four things, and most published ones show none of them: the offer the stranger was replying to, how the list was built and how wide it was allowed to be, the shape of the copy line by line, and the full funnel with a timeframe on it. Emails sent, replies, interested replies, booked calls. A calendar screenshot is not a case study.
How many emails does it take to book 40+ calls in a month?
The e-commerce exit campaign ran about 1,000 emails a day for 30 days and booked 42 calls. The M&A campaign sent just under 40,000 emails to 20,000 prospects over 45 days, two emails each, and booked 47. Volume that size only works if deliverability holds, so the infrastructure gets built and warmed before anything launches.
What reply rate is normal for cold email?
Across these three: a little over 2% for the funding offer, just under 4% for the M&A offer, and a 2% planning target on the e-commerce one. The share of replies that are interested matters more than the raw rate. Those ran near 15% and near 21%, and that is what decides how many calls land.
Do you need heavy personalization if the offer is good?
No. The funding campaign tested a broad relevance line against an AI-written line specific to each prospect's project, and once the offer was performance-based the specific version stopped earning its cost. The M&A campaign used one token, the prospect's industry, read off their website. List precision plus a real offer does the work personalization takes credit for.
How do you keep a wide list from filling the calendar with bad calls?
Put the qualification behind the reply instead of in front of the send. Both exit-focused campaigns used a pre-call form: revenue, profit, inventory and debt on one, financials and exit timing on the other. Anyone who did not clear went into nurture rather than onto the calendar, so volume bought reach without buying wasted calls.
If you would rather have it run for you, the offer, the lists, the infrastructure and someone working the replies, that is what we build and where you can book a call: 3,000+ sales calls booked for clients in the past year. Or keep learning free. Every training sits on the trainings hub, and the full video for the first teardown is on YouTube.
PS - the e-commerce campaign began as a Clay build estimating revenue from website traffic and average order value. It got deleted. The version that booked 42 calls in 30 days checked one thing: is this an e-commerce store.