Cold email benchmarks
Hard numbers we use when planning cold outreach infrastructure. Drawn from thousands of campaigns Allston Labs has run for clients and from publicly-reported data across major sending platforms.
Sending capacity
The hardest constraint in cold email is the per-inbox send cap. Gmail, Outlook, and most ESPs flag any inbox sending more than 30–50 cold emails per day as bulk. Above that, you start landing in spam folders and burning your domain reputation.
| Metric | Safe | Aggressive | Burn zone |
|---|---|---|---|
| Sends per inbox per day | 20–30 | 30–50 | 50+ |
| Inboxes per domain | 2–3 | 3–5 | 5+ |
| New-domain ramp (week 1) | 5/day | 10/day | 30+/day |
| Bounce rate | <2% | 2–3% | 3%+ |
| Spam complaint rate | <0.1% | 0.1–0.3% | 0.3%+ |
Engagement rates by industry
Reply rates vary by ICP, vertical, and personalization quality. These are the bands we see most often in B2B SaaS cold outreach.
| Touch | Open rate | Reply rate | Notes |
|---|---|---|---|
| 1 (initial) | 40–60% | 0.5–1.5% | Highest open, lowest reply |
| 2 (follow-up) | 35–50% | 0.8–2.0% | Where most opens convert |
| 3 (value-add) | 30–45% | 0.5–1.5% | Drop or pivot angle |
| 4–5 (breakup) | 25–40% | 0.3–1.0% | “Should I close the loop?” |
| 6–7 (long-tail) | 20–35% | 0.2–0.5% | Diminishing returns |
| Aggregate | — | 2–5% | Across full 5–7 touch sequence |
Conversion: reply → meeting → opportunity
| Stage | Conversion rate |
|---|---|
| Reply → positive reply | 30–50% |
| Positive reply → meeting booked | 50–70% |
| Meeting booked → meeting held | 70–85% |
| Meeting held → opportunity | 20–40% |
| Opportunity → closed-won | 15–30% |
End-to-end, a healthy B2B cold outreach campaign converts 0.1–0.3% of contacts to closed deals. That means 1,000 contacts → 1–3 deals.
Warmup ramp
New inboxes start at near-zero reputation. You drip-warm them for 2–3 weeks before any real cold outreach, gradually increasing volume so providers learn the inbox sends “normal-looking” mail.
| Week | Sends/day | Reply rate target |
|---|---|---|
| 1 | 1–5 | 30%+ |
| 2 | 5–15 | 25%+ |
| 3 | 15–30 | 20%+ |
| 4+ (steady state) | 30 | 10%+ |
Domain & inbox economics
A standard cold-outreach infrastructure spend, per domain per month:
- Domain registration: ~$12/year ($1/mo) via Namecheap, Cloudflare, Porkbun
- Google Workspace mailbox: $6/inbox/mo (standard)
- Microsoft 365 mailbox: $4/inbox/mo (Exchange Online Plan 1)
- Warmup tool: $20–80/mo per inbox (Mailwarm, Warmup Inbox, Instantly, Smartlead)
- Sending platform: $50–150/mo flat or per-seat (Instantly, Smartlead, Lemlist, Apollo)
A typical 6-domain / 18-inbox setup runs ~$150–300/mo to operate steady-state, plus 3 weeks of warmup before launch.
Deliverability by provider
| Receiver | Inbox % | Notes |
|---|---|---|
| Gmail / Google Workspace | 75–90% | Most forgiving, but enforces auth strictly |
| Outlook.com / Hotmail | 40–65% | Aggressive spam filter, slow to recover |
| Office 365 / Exchange Online | 50–75% | Tenant-level rules vary widely |
| Yahoo / AOL | 60–75% | Stricter than Gmail, looser than Outlook |
| Custom/IT-managed | 30–80% | Hardest to predict; depends on internal gateway |
List quality matters more than volume
The single biggest lever on cold email outcomes is list quality. A verified, ICP-tight list of 500 outperforms an unverified list of 5,000 every time.
- Use double-verification (NeverBounce + ZeroBounce, or similar) — single-tool verification has 5–10% false negatives
- Catch-all domains: send sparingly, treat as 50% deliverable
- Role accounts (info@, hello@): exclude entirely from cold outreach
- Suppress: anyone who replied, opted out, or bounced in the last 90 days
Sources
Compiled from internal Allston Labs campaign data plus public benchmarks from Instantly, Smartlead, Apollo, Lemlist, Mailshake, Reply.io, and Woodpecker (2024–2026 reports).
14 RFC-grounded chapters on the authentication, isolation, warmup, and monitoring stack behind every benchmark on this page.
Open the setup reference →Allston Labs runs cold outreach end-to-end — domain provisioning, inbox warmup, list verification, multi-touch sequencing, reply handling. Forward-deployed engineering by ex-McKinsey + YC operators.
Partner with us →