Cold email warm-up guide

What warm-up is actually for

A new mailbox has no history. Receiving providers cannot tell whether it belongs to a person who sends ten emails a day or a script that will send ten thousand, so they treat the first weeks cautiously. Warm-up replaces that uncertainty with an observable pattern: steady volume, real replies, no complaints.

It is not a switch that a service flips. Automated warm-up networks that exchange messages between mailboxes create a pattern of machine-to-machine mail, which some providers recognise. Real correspondence with real people is stronger evidence, and it is also what you will be doing later.

• Steady daily volume matters more than a high total.
• Replies and non-bulk mail are stronger signals than automated exchanges.
• A warm mailbox is not immune to content or list problems later.

Warm-up sets the pace you can later use; daily sending limits explains how the cap and the interval interact.

A ramp that respects the provider

The exact numbers matter less than the shape. Begin at a level you could sustain for a week without watching it, then raise in steps with a full day between them. If a step produces a block, a deferral storm or a complaint, drop back and hold.

Do not raise twice in one day because the first raise looked fine. Reputation signals are delayed; a burst that looks clean for two hours can produce a block overnight. One change per day is slow enough to attribute the outcome to it.

• Day one: a small number, well within what a person would send.
• Raise by a modest step, never double, and hold for a day.
• Keep the interval between messages at 30 to 60 seconds throughout.
• Pause on any 5xx that names the mailbox or the IP.

The signals that decide whether the ramp worked are covered in sender reputation.

What to send while warming

Send the kind of mail the mailbox will send later, to a small group that expects it. A short note to a colleague, a reply to an existing thread, a question to a supplier: these produce replies and look like correspondence because they are correspondence.

If you must use a warm-up service, keep the real conversations separate and do not let the service set the volume you actually send. The service is a background hum; the campaign is the signal you are trying to establish.

• Real recipients who may reply.
• No links or images in the first messages if you can avoid them.
• Consistent From name and signature from day one.
• No bulk-looking subjects, no promotional language.

When to stop warming and start campaign work

You are ready when the mailbox has a history of steady sending, authentication passes on real messages, and a test of your actual campaign content to a small internal group produced no blocks. That is a factual test, not a number of days.

Worked example: a mailbox warmed for nine days reached 120 messages a day with no blocks. The first campaign batch went to 100 verified recipients at 45-second spacing. It produced two replies and one soft bounce, no hard bounces, and the next day's batch was raised to 150. Nothing about the ramp was heroic, which is the point.

If building the sending layer is the bottleneck, SMTP accounts with warm-up supplies mailboxes that are already configured for outreach.

Reading the first week without overreacting

New mailboxes produce noise. A handful of deferrals in the first days is normal, especially from providers that delay unknown senders. The signal to act on is a rejection that names the mailbox or the IP, or a deferral pattern that persists at the same volume for more than a day.

Do not respond to a single 4xx by dropping to zero. That removes the history you are trying to build. Hold the level, watch the next batch, and only step back if the pattern repeats.

Keep a short log for the week: date, messages sent, accepted, deferred, rejected, replies. It is the only record that later tells you whether the ramp was too fast or the content was the problem.

• A few deferrals early on are expected.
• A 5xx naming the mailbox or IP is a stop signal.
• Hold the level rather than falling silent.

A one-week example ramp and the decision at each step

Example only. The right numbers depend on the provider and the mailbox history, not on this table.

Frequently asked questions