Cold email math: what 1,000 sends honestly turns into
Work the funnel backwards — sends, deliveries, real opens, replies, meetings — with honest numbers at every stage, and see why list quality beats raw volume.
By Norbelys Chirinos, Co-founder
Founder-reviewed ·How we research and correct articles
Every founder asks the same question before their first campaign: how many emails do I need to send to get a meeting? Most answers you’ll find are either vendor fantasy (“book 40 meetings a month on autopilot!”) or so hedged they’re useless. Let’s just do the math, honestly.
The funnel, stage by stage
Say you send 1,000 cold emails to a decent list. Here’s what typically survives each stage:
Delivery — expect to keep 95–98%. Verified, deduplicated lists bounce under 2–3%. If you’re bouncing more, stop: every hard bounce is a reputation hit, and Gmail and Microsoft are watching. 1,000 sends → ~970 delivered.
Real opens — expect 30–50% of delivered. Note the word real. Tools that count Apple’s auto-fetches and security scanners will happily report 60–70% here; humans are fewer. Subject lines and sender reputation move this number. 970 delivered → ~390 humans actually looked.
Replies — expect 2–10% of delivered. This is the widest range in the funnel because it’s where targeting and copy live. Generic blast at a bought list: 1–2%. Tight segment, personalized first line, clear single ask: 8–10%. Call it a healthy 5%: 970 delivered → ~49 replies.
Positive replies — expect 30–50% of replies. The rest are polite no’s, not-nows and out-of-offices. ~49 replies → ~18 interested.
Meetings booked — expect half to two-thirds of interested. People ghost, calendars clash. ~18 interested → ~10 meetings.
So: 1,000 honest sends ≈ 10 meetings
Roughly one meeting per hundred sends with solid execution — and that number is the most honest benchmark we can give you. A great campaign to a hand-built list can double it. A lazy blast can divide it by five.
Why the reply stage swings 5×
The delivery stage barely moves between a good sender and a bad one — 95% versus 98% is a rounding error. The reply stage is the opposite: the same 1,000 sends can return 10 replies or 100, and the gap comes down to a handful of variables stacking on top of each other, not one silver bullet.
- List specificity. “VP Sales at Series B SaaS companies that just raised” replies at a different rate than “anyone with a LinkedIn title containing ‘sales.’” The narrower the segment, the more the first line can sound like it was written for one person, because it was.
- Personalization depth. A first line referencing something true and
recent (a launch, a hire, a specific page they publish) reads as
research. A merge-tagged
{firstName}reads as software. - Offer clarity. “Worth a look?” after one clear problem statement beats a paragraph of feature claims — see what actually makes people reply for the anatomy.
- Industry and role. A founder’s inbox and a mid-level ops manager’s inbox run on different reply norms; benchmarks that don’t segment by audience are averaging across very different behaviors.
- Timing and cadence. A single send gets one shot; a real follow-up sequence recovers replies from people who simply hadn’t gotten to it yet — often a third or more of total replies land on touch two or three, not touch one.
None of these variables is exotic. They’re also the reason two people running “the same” cold email playbook get wildly different outcomes — one of them is quietly doing all five well, and the other is doing one and assuming volume covers the rest.
Two campaigns, same 1,000 sends
The abstract funnel above hides how differently two real campaigns can land with the identical starting number. Here’s a side-by-side of a lazy blast against a tight, well-targeted send — both starting from 1,000 recipients:
| Stage | Blast to a bought list | Tight, verified segment |
|---|---|---|
| Delivered | ~850 (bounce-heavy, unverified) | ~980 (verified beforehand) |
| Real opens | ~255 (30%) | ~490 (50%) |
| Replies | ~17 (2%) | ~78 (8%) |
| Positive replies | ~5 (30% of replies) | ~35 (45% of replies) |
| Meetings booked | ~2–3 | ~20–23 |
Notice where the gap actually opens: not at delivery, where both campaigns land in roughly the same neighborhood, but at every stage downstream of it. That’s the tell that volume was never the lever — targeting and copy were, the entire time, and delivery was only ever the price of admission to a funnel that either works or doesn’t.
Same channel, same volume, an order of magnitude apart in outcome — and the blast also did more damage on the way: a higher bounce rate erodes sender reputation for the next campaign too, so the gap compounds over time instead of resetting each send. That’s the real cost of “just send more”: it isn’t neutral, it’s actively regressive to future performance.
Why this math changes your behavior
Once you see the funnel, the leverage is obvious:
- The reply stage has 5× range; the delivery stage has 1.05×. A day spent improving targeting and your first line is worth more than a week of deliverability tricks — as long as the deliverability floor is solid. Check yours with our free domain health checker.
- Volume multiplies whatever you are. If your 1,000 sends produce 2 replies, sending 10,000 produces 20 replies and a burned domain. Fix the ratio first, then scale.
- You can’t improve what’s miscounted. If your tool reports 64% opens and they’re 41% real, every conclusion downstream is wrong. Decide on replies — the number no bot can inflate.
Working it backwards
Need 20 meetings a month? At honest averages that’s ~2,000 well-targeted sends — about 100 per working day, comfortably within what a couple of healthy mailboxes can do with human pacing. That’s the whole plan: small daily volume, clean list, sharp copy, honest measurement, repeat.
Two quick answers
Does a bigger list mean more meetings? Only if delivery, opens and replies hold their rate as the list grows — and for most senders they don’t, because a bigger list is usually a less targeted one. Ten thousand loosely-matched sends at a 1% reply rate produce fewer meetings than 1,000 tightly-matched sends at 8%, and burn a domain doing it.
How long before the funnel stabilizes enough to trust these numbers? Give it at least two full sequences (roughly 300–500 sends through every step, not just the first touch) before drawing firm conclusions. Early results skew heavily on whichever variant or segment happened to go out first; the funnel needs enough volume per stage to stop being noise and start being signal you can actually act on.
No autopilot fantasy. Just math that holds up.