Most cold email programs don't fail because of bad copy. They fail because someone tried to push 400 emails a day through two mailboxes, watched placement collapse in week three, and blamed the offer. Infrastructure is arithmetic, and the arithmetic is rarely done before launch.
This guide gives you a working formula for sizing your sending infrastructure: how many mailboxes, how many domains, and how much daily volume is safe. It works backward from the number that matters, which is meetings booked, so you buy only the capacity you need.
Work backward through the funnel. Pick a monthly meeting goal, then divide by your conversion rates.
Example: you want 20 qualified meetings per month. Using conservative planning numbers:
- Positive reply to meeting booked: 50% - Positive reply rate (of contacted prospects): 1.5% - Meetings needed: 20, so positive replies needed: 40 - Prospects to contact: 40 ÷ 0.015 ≈ 2,670 new prospects per month
These are planning assumptions, not guarantees. Replace them with your own numbers after the first 500 sends. Our cold outreach metrics guide explains which benchmarks to track and how to read them.
Each prospect receives a sequence, typically three to four touches. So 2,670 prospects at 3 touches is roughly 8,000 emails per month, or about 380 sends per business day (21 working days). That is the number you size infrastructure against.
Takeaway: compute required daily sends from your meeting goal before you touch a domain registrar.
Every mailbox has a safe daily ceiling, and exceeding it is the fastest way to damage reputation. A defensible rule for cold sending:
- Fully warmed mailbox: 30 to 40 cold emails per day, plus a small amount of warmup traffic - Mailbox in weeks 1 to 3: 5 to 15 per day, ramping gradually (see our warmup playbook) - Hard limit: never approach the provider's published sending maximum. That is a technical cap, not a safe operating target
At 35 sends per mailbox per day, our 380-send target requires about 11 active mailboxes. Add 20% headroom for mailboxes that are resting, ramping, or paused, and you need roughly 13 to 14.
Why not just send 150 per mailbox? Mailbox providers evaluate sending patterns per sender. A new address that suddenly sends at volume looks like compromised or abusive behavior. Spreading volume keeps each sender's footprint small and human-like, which is the core of every deliverability audit we run.
Takeaway: active mailboxes = daily sends ÷ 35, plus 20% headroom.
Mailboxes cluster under domains, and domains carry reputation. Concentrating too many senders on one domain means a single problem takes everything down with it.
A practical structure:
- Two to three mailboxes per sending domain. For 14 mailboxes, that is five to seven domains - Never send cold email from your primary domain. Use close variants (for example, a different TLD or a prefix/suffix of your brand) that redirect to your main website - Reserve one or two spare domains in warmup at all times so you can rotate in replacements without a gap in volume - Configure authentication on every domain before the first send: SPF, DKIM, and DMARC. Our SPF, DKIM, and DMARC setup guide walks through each record
The spare-domain rule is the one teams skip. When a domain's placement dips, you want to pause it for recovery and shift volume immediately, not scramble to buy and warm a replacement over the next three weeks. If a domain is already damaged, see how to recover a burned domain.
Takeaway: domains = mailboxes ÷ 2.5, plus two warming spares.
Using the same planning assumptions (1.5% positive reply rate, 50% reply-to-meeting, 3-touch sequences, 21 sending days):
| Monthly meetings | New prospects | Daily sends | Active mailboxes | Domains (incl. spares) | |---|---|---|---|---| | 10 | ~1,330 | ~190 | 6-7 | 4 | | 20 | ~2,670 | ~380 | 13-14 | 7-8 | | 40 | ~5,330 | ~760 | 26-27 | 12-13 | | 80 | ~10,670 | ~1,520 | 52-53 | 23-25 |
If your reply rate is better than 1.5%, you need less. A tightly targeted list with relevant personalization can materially cut required volume. That is why we put so much weight on prospect qualification: scoring leads against a clear ideal customer profile means fewer sends to reach the same meeting count, which lowers risk at every layer of the stack.
Takeaway: better targeting is cheaper than more mailboxes.
Having capacity is not enough. How you distribute volume across it determines whether it holds up.
1. Rotate senders per prospect, not per email. Within a sequence, every touch to a given prospect should come from the same mailbox. A thread that changes sender mid-sequence looks wrong to recipients and filters alike. 2. Randomize send times inside a window. Fifty emails at exactly 9:00:00 is a pattern. Spread them across business hours in the recipient's time zone. 3. Cap per-domain volume, not just per-mailbox. If three mailboxes share a domain, the domain's total should stay under roughly 100 per day. 4. Throttle automatically on warning signals. Bounce rate above 2%, or any spike in spam complaints, should reduce a mailbox's quota automatically, not wait for a weekly review. 5. Keep sequences short. Extra follow-ups add volume and complaint risk for diminishing returns. Our follow-up email strategy covers where the return curve flattens.
Our platform handles this rotation, throttling, and health-based pausing automatically: it assigns each prospect a sender, keeps the whole sequence on that sender, and reduces a mailbox's quota when bounce or complaint thresholds are crossed. If you run it manually, put these rules in a spreadsheet and review them daily.
A rough monthly infrastructure budget for the 20-meeting plan: 8 domains at about $1 per month amortized, 14 mailboxes at $6 to $12 each, plus a verified data source and sending platform. Infrastructure typically lands in the low hundreds of dollars per month before data and software.
Compare that with a fully loaded SDR, which commonly runs well over $100,000 a year once salary, benefits, tooling, and management time are counted. Automated prospecting does not replace the judgment of a strong closer, and we recommend a hybrid model where humans handle live conversations. But for top-of-funnel volume, the cost gap is large. For a deeper breakdown, read our analysis of cost per meeting.
- Buying for peak, warming for never. Mailboxes bought on Monday cannot send at volume on Friday. Plan a three to four week lead time. - Ignoring list quality. Sending 380 emails a day to an unverified list produces bounces that burn every mailbox you own. Verify first (see our guide to building verified B2B lists). - Scaling too fast. Increase total volume no more than 20 to 30% per week and watch placement before each step. More on this in scaling without killing deliverability. - No monitoring. Without per-domain bounce, reply, and placement tracking, you find out about a problem when meetings stop, weeks late.
Sizing cold email infrastructure takes five minutes of math: meeting goal, divided by conversion rates, gives required sends; sends divided by 35 gives mailboxes; mailboxes divided by 2.5 gives domains; add headroom and spares. Then enforce rotation and throttling rules so the capacity you bought actually holds up.
If you'd rather not manage warmup schedules, sender assignment, and health monitoring by hand, OnyxSend handles that layer automatically, so your team can focus on the conversations that follow. You can see how it works and get started on our pricing page.