What the first 90 days of an outbound programme actually contain

A founder signs an outbound engagement in the first week of March and asks when the meetings start. The honest answer is late April, and the reason is that three weeks of it belongs to a mail server rather than to anybody's effort.

This is what the first 90 days actually contain, week by week, so you can hold the plan against a calendar instead of a hope.

Weeks 1 and 2: decide what you are testing

No sending happens. Two things get built in parallel.

The offer test. Four questions answered in writing with the person who closes deals: who signs by function, what breaks without you in their units, what proves the claim, and why this quarter. An hour of conversation, then a document. If the answers are placeholders, the campaign will not diagnose itself later and the honest move is ten customer conversations first.

The list definition. The universe as a query against a real source, the disqualifiers, and the dated trigger. This produces an account count, and the account count is the first checkpoint. Under 200 is too narrow to learn from. Over 5,000 means the trigger is doing no work.

In parallel, the boring part starts: domains registered, DNS records published, mailboxes created. It has to happen in week one because of what week three needs.

Weeks 3 and 4: build the list and write

Warm-up continues without you. The work moves to research.

Pull the universe from the register. Apply disqualifiers. Layer the trigger and keep only accounts where the event happened inside the window. Then, and only then, resolve contacts, because enrichment is the expensive step and the volume has just dropped by 90 percent.

Verify every address. Segment anything under 90 percent confidence into its own low-volume stream. Record the source per row, which you will need the first time somebody asks where you got their address.

Copy gets written against segments rather than against the whole list. Four variants across four segments beats one message to everyone, because the comparison is what produces learning in week eight.

Weeks 5 and 6: send at low volume

The first real sends go out. Volume stays deliberately low, around 20 to 30 per mailbox per day, and the point of these two weeks is not meetings.

You are checking four things: does the mail reach an inbox rather than Junk, do the addresses bounce, does anybody reply at all, and do the replies say what you expected. Seed mailboxes at Google, Microsoft and one European provider tell you the first one, and your own sending tool cannot.

Expect the first meetings here. Two or three from a few hundred sends. They are not the result, they are evidence the machine turns.

The single most common way this phase goes wrong: somebody decides the volume is too low, pushes it to 100 per mailbox, and burns a domain that was two weeks from being an asset. Ramp is not a suggestion.

Weeks 7 to 10: full volume and the first read

Volume reaches plan, somewhere near 30 sends per mailbox per day. For a three-domain, nine-mailbox setup that is roughly 5,000 first touches a month once follow-ups are counted.

Around week eight you have 1,500 sends per segment, which is the point where the numbers stop being noise. Read them in order and stop at the first one out of range.

This is also when the negative replies become useful. A third of your replies are no's, and half of those contain a reason. Coded by segment, they tell you which country to drop, which title actually answers and which objection repeats.

Weeks 11 to 13: cut and concentrate

The second version of the campaign is built from the first one's evidence. Segments that produced nothing get dropped rather than optimised. The message that won gets carried into the segments that had a weaker variant. The ICP document is rewritten with what the market argued with.

Expect the meeting count to roughly double between month two and month three, from the same volume. That improvement comes from cutting, not from sending more.

What the 90 days should have produced

Twelve to twenty-seven meetings across the quarter, and a documented answer to whether this market deserves the next four. The second output is worth more than the first, and it is the one that gets left out of the proposal.

The three ways this timeline breaks

Sending before the warm-up finishes. Costs the domain and restarts the clock.

Changing the message every ten days. Guarantees no version accumulates enough sends to be readable, so month three has the same questions as month one.

Nobody owning the inbox. A reply answered within the hour books at twice the rate of one answered tomorrow. If the founder is the only person who can answer and the founder travels, the programme loses a third of its output to a calendar.

Agree the review point before the first send. Fifteen hundred sends per segment or eight weeks, whichever comes first, and then hold to it while it is uncomfortable. That is the whole discipline.