Outbound infrastructure · sizing tool

Cold email capacity calculator

Set the question you want answered and the time you'll give it. The tool works forward to contacts, emails and mailboxes — and shows what each additional inbox actually buys you. Everything is judged on conversion — a booked meeting or a closed sale. A probe takes as long as its sequence takes; inboxes buy concurrency, not speed.

What are you asking?
Metric
Fixed — meeting booked rate Qualified meetings (or sales) ÷ unique prospects contacted Per prospect, never per email. A 3-email sequence to 1,000 people is 2,850 sends but 1,000 trials.
1.00%

The rate you expect from a typical or dead audience.

2.0× → 2.00%

The smallest lift worth acting on. Everything below this you accept you'll miss.

Scope
18
Sequence — this sets the clock
3 → 2.85 sends
3 days
12 days
One probe takes 20 days sends on day 0 · 3 · 8, then a 12-day reply tail
4 → 80 days

Mailboxes buy concurrency, not speed. One probe always takes a full cycle; more inboxes mean more cells running side by side inside it.

Constraints & prices
25

25 is the deliverability ceiling from deck 5. Raising it is a reputation decision, not a capacity one.

25,000

Every name can only be probed once. This is usually what actually binds.

$0.18
Contacts—people, total
Emails—sends, total
Mailboxes—to finish in time
——
Per arm / cell—
Expected hits / cell—
Cells in flight—
Calendar—
Emails / month—
Mailbox $/mo—
Contact data $—
ICP consumed—

—

—

What an inbox buys you — and what a wave buys you

Smallest effect detectable in the chosen window, as you add mailboxes. The curve flattens fast — that flattening is the argument against buying more.

The same question at other horizons

Mailboxes needed to answer this exact question in each window, and what the current plan's capacity could detect instead if you kept the mailbox count fixed.

Spread overMailboxesIn flight$ / moFeasible on list

How it computes

  • Compare — two-proportion z-test, α=.05 two-sided, 80% power, per arm. Reproduces deck 3's "~2,500 for 3%→4.5%" (2,518) and deck 4's "~1,150 for 2%→4%" (1,141).
  • Screen — one-sided single-arm test of the baseline against the live threshold, α=.05, 80% power. Cheap by design: this is the metal detector, not the assay.
  • Contacts → emails — 1.00 / 1.96 / 2.85 / 3.70 / 4.50 sends per contact for sequences of 1–5, derived from ~8% cumulative reply and ~2% bounce attrition. Not a published figure.
  • One denominator: unique prospects contacted. The metric is meetings booked (or sales closed) ÷ prospects, and every rate on this page uses that base. This is forced by the test, not a preference — randomisation is at the person level, so a prospect who receives three emails is one trial. Published per-prospect bands: below-average <0.4%, average 0.4–0.8%, above-average 0.8–1.5%, top quartile 1.5–3.0%.
  • Vendor figures quoted per send are converted, not discarded. Several sources define reply and meeting rates as "÷ total emails delivered", which is why published benchmarks span an implausible range — the spread is denominators, not performance. Deck 4's Gong figure of 0.29% per email is 344 emails per meeting; at 2.85 sends per prospect that is 0.83% per prospect, which lands in the above-average band rather than looking like an outlier. Woodpecker's per-prospect 8.3% reply rate for 3–5 step sequences reconciles the same way against Saleshandy's 3.4% per delivered email. Both decks are consistent with the bands above once put in the same units.
  • Cycle length — derived, not chosen. Send gaps widen by 2 days each step, so the locked 3-email sequence at a 3-day first gap lands on days 0 · 3 · 8; add the reply tail for 20 days to verdict. Buying inboxes cannot compress this.
  • Mailbox capacity — sends/day × the working days inside a cycle (5/7 of calendar). Warm-up is not modelled: a new mailbox delivers roughly 35% of this in its first month, so add a cycle before the clock starts.
  • Not modelled — multiple-comparison correction (screening 18 cells at α=.05 yields ~1 false positive per pass), segments too small to fill an arm, and reply-handling labour.