Quick answer
Plan for 60 to 90 days before you trust a cold email number as your real number. Domain warmup takes 2 to 4 weeks before you can send real volume, first replies start trickling in around week 5 or 6, and the sample size needed for a fair read on reply rate usually isn't there until month 2 or 3.
The short version
I'm Hlib Storchak. I build outbound systems for B2B founders and sales teams, and most of what follows comes from running this for clients, watching the same timeline repeat almost identically across accounts that have nothing else in common. If you're asking how long cold email takes to work, you're really asking two different questions that get conflated: how long until it's technically running, and how long until the number it produces is one you can trust. The first answer is roughly a month. The second is roughly a quarter.
That gap is the part almost nobody plans for, and it's the reason so many programs get killed in week three for looking like a failure when they were never far enough along to be judged at all.
Weeks 1-2: domain and infrastructure setup
Nothing you'd call a cold email campaign happens yet. This is domain purchase, DNS records (SPF, DKIM, DMARC), mailbox creation, and the start of automated warmup, where a tool sends and replies to small volumes of email between seeded inboxes to build sender reputation before a single prospect sees anything. I've written the day-by-day mechanics separately in how to warm up cold email domains, so I won't repeat the steps here, but the timeline matters: most operators treat day 14 to day 21 as the earliest a brand-new domain is trusted enough for meaningful prospect volume, not day 1.
Skipping or shortening this stage is the single most common reason a program looks broken in month one. It isn't broken, it's just not warmed up yet, and sending real volume into an unwarmed domain is what actually breaks it.
Tip. Buy 2 to 4 sending domains per brand domain, not one. A single domain carrying all your volume means a deliverability problem is also a total outage. Spreading volume across a few domains, each individually warmed, is standard practice, not overkill.
Weeks 3-4: first sends, not first signal
Real prospect emails start going out, usually at a deliberately low daily volume per mailbox (20 to 40, not 200) while the domain's reputation is still young. Opens start showing up. Replies mostly don't yet, or arrive in single digits. This is normal, not a sign the copy is wrong, and it's the stage where I see the most premature copy rewrites: someone reads three days of an 8% open rate and starts rewriting subject lines that were never given enough sends to mean anything.
The one number worth watching closely this early is bounce rate, not reply rate. A bounce rate above 2% at this stage is a genuine early warning that the list or the domain has a problem, and it's the one metric that's meaningful before volume is high enough for anything else to be.
Weeks 5-6: replies start, in small numbers
By now sending volume is closer to a steady state and the domain has enough history that replies start arriving at something like the rate the account will eventually settle at, just on too small a sample to trust yet. This is the stage where founders and reps start forming an opinion, and it's also the stage where that opinion is least reliable: a good week and a bad week both look dramatic when the total send count is still in the hundreds, not thousands.
This is the setup I run for clients: hold the copy and targeting steady through this stretch unless the bounce rate itself flags a problem, and resist judging the reply rate until the volume behind it is large enough for a single good or bad week to stop swinging the average.
Weeks 7-12: the ramp to a trustworthy number
Sending volume reaches its intended steady state, the domain's reputation is established, and enough emails have gone out that the reply rate stops moving much week to week. This is also, separately, when the human side of the ramp catches up: a founder or a new rep running their own outbound gets measurably better at the job over the same stretch, which is its own reason not to judge month one against month three.
Somewhere in this window is when I tell clients they finally have a number worth acting on, whether that's "this offer works, scale it" or "this offer needs to change." Before that, every decision is being made on a sample too small to separate signal from noise.
Timeline at a glance
| Stage | Timeframe | What's actually happening | What you can conclude |
|---|---|---|---|
| Domain and infra setup | Weeks 1-2 | DNS records, mailbox creation, warmup begins | Nothing yet, this is infrastructure, not a campaign |
| Low-volume first sends | Weeks 3-4 | Real prospects, low daily volume, opens appear | Watch bounce rate only, ignore reply rate |
| Replies begin | Weeks 5-6 | Small numbers of replies, sample still thin | A trend is forming, not yet reliable |
| Full ramp | Weeks 7-12 | Steady-state volume, established domain reputation | Reply rate is now a real number to act on |
| Mature program | Month 3+ | Rep and offer both fully ramped | Fair to compare against published benchmarks |
Why it genuinely takes this long
Two separate ramps are running at once, and both are slow for structural reasons, not because anyone is doing it badly. The domain ramp exists because mailbox providers deliberately throttle and watch new sending domains, since that's exactly the profile a spammer also has, and there's no legitimate way to skip the trust-building period, only ways to shorten it slightly with a clean setup. The human ramp exists because writing and adjusting outbound copy, reading replies, and calibrating an offer is a skill that improves with repetitions, and a founder or rep sending their first cold email campaign is, by definition, on their first repetitions.
The Bridge Group's SDR benchmark research puts the average human ramp to full productivity at around three months, which lines up closely with what I see on the domain and infrastructure side too. I've gone deeper on what a mature number should actually look like once you get there in realistic cold email benchmarks. Comparing a two-week number against a published benchmark is comparing an unramped program to a fully ramped one, and it's the single most common way people talk themselves out of a channel that was actually working.
The mistakes that stretch the timeline further
Most of the delays I see aren't inherent to cold email, they're self-inflicted, and they're avoidable once you know to watch for them.
- Sending full volume from day one. This doesn't shorten the warmup, it usually damages the domain's reputation and adds weeks of recovery on top of the timeline you'd have had anyway.
- Rewriting copy every few days based on early numbers. Each rewrite resets the sample you're judging it on, so you never accumulate enough sends on any single version to actually learn anything.
- Running one domain for everything. A single deliverability incident takes the whole program offline instead of one slice of it, and recovering a damaged domain's reputation takes longer than warming a fresh one.
- Skipping list quality to hit a volume target. A rushed, loosely matched list inflates bounce rate, which is the one metric that actually damages the domain during the exact window it's trying to build trust.
- Treating week 2's reply rate as the verdict. It's the most common reason I see a genuinely working channel get killed before it had a fair chance to show it.
Can you speed any of this up?
Partially, and it's worth being honest about the limits. Starting with domains that have some sending history (rather than brand new registrations) can shave a bit off the trust-building window, and a tight, well-matched list reduces the risk of a bounce-rate problem that would otherwise add weeks of recovery. Running more domains in parallel from day one increases total volume sooner without needing any single domain to carry more than it safely can.
What doesn't shorten it: skipping the warmup step, sending volume ahead of what the domain has earned, or trying to judge results faster than the sample size allows. Those don't compress the timeline, they just add a deliverability problem or a bad decision on top of the timeline you already had.
The real question isn't speed, it's when to judge
The mistake I see most often when I take over an account isn't a bad offer or weak copy, it's a program that was judged, and sometimes killed, a month before it had produced enough volume to be judged fairly. If you take one thing from this article, take the discipline of picking a judgment date, roughly 60 to 90 days out, and holding to it regardless of how week three looks. A program that's genuinely broken will still look broken at day 75. One that just needed time won't.
Key takeaways
- Setup (weeks 1-2) and signal (weeks 7-12+) are different things. Don't judge the first as if it were the second.
- Bounce rate is the only metric worth watching in the first few weeks. Reply rate needs volume before it means anything.
- The domain ramp (2 to 4 weeks) and the human ramp (roughly 3 months, per Bridge Group data) run in parallel, and both need to finish before a number is trustworthy.
- Full volume on day one doesn't speed things up, it usually causes a deliverability problem that adds weeks.
- Pick a 60 to 90 day judgment date before you start, and hold to it.
Does the platform you pick change the timeline
The underlying timeline (domain trust, ramp, sample size) doesn't change based on software. What changes is how much manual work the warmup and sending steps take off your plate. For sending and warmup itself, Salesforge and Warmforge are what I default to for clients, mostly because the warmup runs on the same infrastructure as the sending rather than as a bolted-on separate tool, but the week-by-week timeline above holds regardless of which platform you pick. If you're comparing options, I've written a direct Salesforge vs Lemlist breakdown that judges both on the merits rather than picking a winner in advance.
FAQ
How long does cold email take to work?
Plan for 60 to 90 days before you trust the number you're seeing. Domain warmup takes 2 to 4 weeks before real volume is safe to send, replies start in small numbers around week 5 or 6, and the sample size needed for a reliable reply rate usually isn't there until month 2 or 3.
Why don't I see replies in the first week?
You shouldn't expect them. The first 1 to 2 weeks are domain and infrastructure setup, and the following 2 weeks are deliberately low-volume sends while the domain builds trust. Replies are a function of volume and time, and neither is there yet.
What should I actually watch in week one?
Bounce rate, not reply rate. Above 2% signals a list or deliverability problem worth fixing immediately. Reply rate on a sample of a few dozen sends isn't a number yet, it's noise.
Can I speed up domain warmup?
Slightly, by starting with a domain that already has some sending history and keeping the list tight enough to avoid a bounce-rate problem. You cannot skip it: sending full volume from an unwarmed domain usually costs more time in recovery than the warmup would have taken.
When should I decide if cold email is working or not?
Not before 60 days, and 90 is safer. A program that's genuinely broken will still look broken at day 75. Judging it at day 15 mostly measures whether the domain is warmed up yet, not whether the channel works.