Meet the program
The outbound program of a US advisory firm: cold email to business owners, run on dedicated infrastructure, never on the firm's main domain.
The challenge
The firm's earlier attempt at cold email had produced nothing, and the reason was the list: pattern-guessed email addresses. One lead file from that era held 298 addresses, and 42 of them were weak-confidence name-pattern guesses sitting on valid domains. Guessed emails bounce, bounces destroy sender reputation, and destroyed reputation lands everything after it in spam.
The rebuild had one rule above all others: the sending base has to survive the campaign. Volume means nothing if the inboxes are dead by week two.
The solution
A pool, not a mailbox
Thirty inboxes were prepared across Microsoft and Google, with about ten held in reserve. Live sending ran on ten Google inboxes, each capped at 25 emails a day, for a combined ceiling of 250 a day. No single inbox ever carries the campaign. On October 3 the per-sender limit was raised to 50 a day, and only after the health record justified it.
Published emails only
The list standard changed permanently: only email addresses actually published on a source page, each with its source recorded. If no real email exists, the contact is skipped. No patterns, no guesses, ever.
Sequences built to protect the senders
Eight steps: one personalised first email and seven in-thread follow-ups, with waits of 2, 3, 3, 4, 4, 5 and 6 days and empty subject lines so replies thread properly. Plain text only. Open and click tracking off, because tracking pixels hurt deliverability. Stop on reply on, so no prospect ever gets chased after answering. Provider matching on, so Google inboxes write to Google addresses and Microsoft to Microsoft where possible.
The results
Verified sends per campaign, read back from the sending tool at the October 1 check: Monday 324, Tuesday 205, Wednesday 251, Thursday 237, Friday 239. The campaigns kept working their follow-ups after that; by the October 3 state check the totals stood at Monday 401, Tuesday 278, Wednesday 433, Thursday 245 and Friday 239.
The Friday campaign is the one to study. Its last 206 sends went out in a special 7 to 10 PM Eastern window, 200 of them landed, and the campaign closed with a single hard bounce, a 0.42% bounce rate. Five leads were never contacted because the window closed first, and one lead was invalid on verification. All ten live senders finished at 100% health, each having sent its full 25 of 25 that day.
We will say the quiet part plainly: reply volume in this first wave was low. That is not what this wave was for. It proved the base: a pool that sends real volume, barely bounces, and stays healthy enough to compound. Most cold email programs die because the base rots. This one did not.
The return on the discipline: 1,596 verified sends across the five campaigns by the October 3 check, one hard bounce in the final campaign, and every sender still at 100% health. That means the next wave starts from a warm base instead of a rebuild. An inbox pool is capital. This one finished the wave worth exactly what it was worth at the start, plus a sending record.
FAQ
Why cap each inbox at 25 emails a day?
Reputation is per inbox. Low daily caps with more inboxes spread risk, keep each sender's pattern close to human, and mean one flagged inbox never takes the campaign down.
Why turn open and click tracking off?
Tracking needs a custom domain and image pixels, both of which spam filters read as bulk signals. Plain text with no tracking trades vanity metrics for inbox placement.
What does provider matching do?
It sends from a Google inbox to Google-hosted addresses and from a Microsoft inbox to Microsoft-hosted ones. Same-provider delivery tends to pass filters more easily.
How long is warmup before a pool like this sends?
Our standard ramp: 14 days of warmup, then 10 sends a day in week one, 20 in week two, and 25 a day from week three, rising only while health stays clean.
