Warming up a domain is the least glamorous and most consequential part of a cold-outreach operation. A bad warm-up shows up as a 20% inbox rate three weeks into a campaign, after you've already paid for tooling and spent a month on copy. A good warm-up is boring, takes a month, and is worth more than every other optimisation combined.
Warm-up is 80% recipient quality, 20% volume curve. Start at 10–20 sends per day to engaged recipients, double every 2–3 days, and aim for a reply rate above 10% throughout. Automated warm-up services help in controlled ways; Gmail now detects several of them.
What you are actually warming up
Three separate reputations, all built at different rates:
- Domain reputation. Per-domain history at Gmail, Outlook, Yahoo and the rest. Built by consistent authenticated sending to engaged recipients over weeks. Persists across sending IPs.
- IP reputation. If you're on a dedicated IP, the IP has its own reputation separately from the domain. Shared IPs share a pooled reputation.
- Engagement signals. Per-domain, not per-sender — Gmail models "is this domain one users want to hear from?". Replies, stars, folder moves all contribute. Opens contribute much less than they used to.
A warm-up that produces volume without engagement — classic automated tool behaviour — moves the first two needles modestly and the third not at all.
Week 1: 10–20 per day, engaged recipients only
Day 1 through day 7. Send 10–20 messages per day, all to recipients you know will open, read and ideally reply. Your own team, co-founders, advisors, friends who agree to help. Every recipient replies. Every recipient stars the message or moves it to Primary if it lands in Promotions.
The goal in week 1 is not volume. It is establishing a pattern that looks like a real person running a business from this domain: reasonable send-times, real content, immediate engagement. Send between 09:00 and 17:00 sender-local. Take weekends off.
Week 2: 30–50 per day, introduce external recipients
Begin including external recipients on Gmail, Outlook and one privacy-first provider (Fastmail or iCloud). Keep the engaged core — friends, advisors — as the majority. External recipients should be people who have consented to receive your outreach during warm-up: other founders, early pilot contacts, anyone who has explicitly agreed. Not cold prospects.
Replies still carry the most weight. Opens help. Movements out of Promotions into Primary are measurable wins — every move logs an engagement signal.
Week 3: 100–200 per day, first real prospects
Now you can start real cold outreach, but the first prospects should be your highest-confidence segment: the contacts most likely to reply to your actual pitch. Not a wide-net blast. Aim to keep reply rate above 8–10% through this week. If it drops, pull back volume and recover.
Watch bounce rate carefully. Any bounce rate above 2% signals list-quality issues that will compound reputation damage fast. Pause, re-validate, resume.
Week 4: 300–500 per day, full campaigns
Only in week 4 do you begin running full cold-outreach volume. Continue distributing across multiple sending inboxes — one inbox sending 500/day is a very different reputation signal from 10 inboxes sending 50 each. Keep rotation even.
From week 4 onward, run weekly placement tests. Any drop in inbox rate below 70% is a signal to pull back volume and diagnose before continuing.
Engagement requirements throughout
Replies are the single most valuable warm-up signal. Gmail treats a reply as strong evidence the recipient wants to hear from you. Opens contribute; clicks contribute; but replies dominate. Target 10% reply rate during warm-up. If you can't generate that from the recipients you have access to, your warm-up is under- engaged and the domain will struggle when you hit real prospects.
The other high-value signals, in rough order:
- Reply by the recipient.
- Move from Promotions / Spam to Primary / Inbox.
- Star, label, or "mark important".
- Click on a link.
- Open (heavily degraded by MPP and bot pre-fetch).
Automated warm-up services
Mailwarm, Warmup Inbox, Lemwarm, Warmy and similar services run networks of mailboxes that exchange messages automatically to generate engagement signals for subscribed domains. What they actually do: send to a network of peer inboxes, reply, open, sometimes mark as important.
Two problems as of 2025. First, Gmail's ML model has gotten much better at detecting these networks. Messages that bounce between a known warm-up pool are increasingly weighted as non-engagement. Lemwarm especially was caught in Gmail's 2024 updates, and inbox-rate gains from it dropped noticeably. Second, warm-up networks produce engagement that doesn't transfer to real prospects. A domain with a great Lemwarm score still faces cold-lead reality on launch day.
Verdict: warm-up services help at the margin. They do not replace genuine recipient engagement. Use them alongside real warm-up, not instead of it.
Manual warm-up: team mailboxes, personal connections
The most effective warm-up we see is genuinely manual. Team mailboxes reply to new sender emails. Personal connections are invited to exchange real messages. Early prospect conversations are routed through the warming domain on purpose. This produces engagement signals that match the exact pattern you will run at scale — because they are the same pattern.
Practical setup: send a genuine business update to a list of 30 engaged contacts on day 1. Ask for a reply. Repeat every 3 days with new content. Add 10 new contacts each week. By week 4 the domain has a real engagement history with real humans.
How to verify warm-up is working
- Weekly placement test across 20+ providers. Inbox rate should rise week over week.
- Google Postmaster Tools — domain reputation should move from "Low" through "Medium" to "High" over 3–4 weeks. If it stays "Low", warm-up is failing.
- Spamhaus and URIBL checks throughout — any listing stops the warm-up until delisted.
- Bounce rate stays under 2% and complaint rate under 0.1%.
Common warm-up mistakes
- Volume bursts. Jumping from 10 to 200 sends in a day undoes the curve.
- No engagement signals. Sending volume without generating replies or folder moves. Warm-up services alone fall in this trap.
- Inconsistent hours. Sending at random times across timezones. Pattern consistency matters.
- Sending to unverified lists during warm-up. A 5% bounce rate in week 2 ruins the warm-up. Only send to verified addresses.
- Skipping weekends arbitrarily, then sending. Pick a sending pattern and keep it consistent.
Warm-up is 80% recipient quality. 1,000 messages to engaged humans who reply and star is worth more than 10,000 messages through a warm-up network. Build the engaged core first; everything else follows.