Before a new email account sends real outreach, it spends a warm-up period sending and receiving practice emails. We tried 7, 14 and 30 days of warm-up, and for half the accounts, the first few days of real emails went to people who had replied positively to us before. The accounts that warmed up longest and started with those people stayed out of spam the longest.
February, cold from day oneMarch, past responders first
We wanted to know two things: whether more warm-up time helps a new Gmail account stay usable for longer, and whether it matters if its first emails go to strangers or to people who have replied positively to us before.
So we built six separate groups. Each one had 300 unique Gmail accounts on 100 unique domains, three accounts per domain. That's 1,800 accounts and 600 domains overall, and no account was used in more than one group.
The three February groups sent cold outreach from their first sending day. The three March groups spent their first 2 or 3 sending days on previously engaged contacts, then switched to normal cold outreach. MailReach monitored all six.
| Group | Started | Warm-up | First sending days | Accounts / domains |
|---|---|---|---|---|
| 1 | February | 7 days | Cold outreach straight away | 300 / 100 |
| 2 | February | 14 days | Cold outreach straight away | 300 / 100 |
| 3 | February | 30 days | Cold outreach straight away | 300 / 100 |
| 4 | March | 7 days | Past positive responders for 2 to 3 days, then cold | 300 / 100 |
| 5 | March | 14 days | Past positive responders for 2 to 3 days, then cold | 300 / 100 |
| 6 | March | 30 days | Past positive responders for 2 to 3 days, then cold | 300 / 100 |
The groups were separate cohorts, so we're comparing different accounts, not the same accounts tested twice. Every account went through the same four steps.
One account's life in the study, at 15 emails a day from its first sending day. Lifespan counts from that first sending day, including the March accounts' 2 or 3 days on past responders, to the first check below 3. Warm-up days aren't included.
We only wanted to change two things: how long an account warmed up, and who got its first emails. The infrastructure stayed the same for every group.
We used Gmail sending accounts only, and only emailed Gmail addresses. Addresses behind email security gateways such as Mimecast and Barracuda were filtered out. The domains and accounts came from an established reseller whose setup had been working for us for months before the test, and we applied the reseller's recommended warm-up settings the same way everywhere.
| Item | Setup |
|---|---|
| Sending accounts | Gmail |
| Recipients | Gmail addresses only, with security-gateway addresses filtered out |
| Sending software | Instantly |
| Domains and accounts | An established reseller |
| Accounts per domain | Exactly 3 |
| Domains per group | Exactly 100 |
| Accounts per group | Exactly 300 |
| Overall | 600 unique domains, 1,800 unique accounts |
| Technical checks | SMTP and sending configuration, DKIM and DMARC, plus other relevant settings |
| Warm-up settings | The reseller's recommended configuration |
Volume stayed fixed too. Every account was held at 15 emails a day, in the cold-first and engaged-first groups alike. That's 45 a day per domain and 4,500 a day per group when all 300 accounts in it are sending. Those are calculated capacities: we're not saying all six groups sent at the same time, or that every day hit the maximum.
| Level | Accounts | Emails a day at 15 per account |
|---|---|---|
| One account | 1 | 15 |
| One domain | 3 | 45 |
| One group | 300 | 4,500 |
| All six groups, if sending at once | 1,800 | 27,000 |
The test ran inside a larger, established outreach operation: more than 5,000 email accounts sending about 100,000 emails a day. Those accounts normally got a 14-day warm-up and went through the same technical and MailReach checks. These numbers describe the operation the test sat inside. They aren't the size of the test.
| Existing operation | Six-group test | |
|---|---|---|
| Email accounts | More than 5,000 | 1,800 unique |
| Daily sending | About 100,000 emails | Up to 4,500 per group (15 per account) |
| Usual warm-up | 14 days | 7, 14 or 30 days |
| Domains | Not specified | 600 unique |
| Purpose | Ongoing outreach | Comparing how long accounts last |
"Died" didn't mean disconnected, unable to send, or suspended by Gmail. It meant deliverability had dropped below the score we treat as usable. We checked every account on MailReach every five days. A score below 3 meant testing showed spam placement, and those accounts were also getting replies at close to 0%.
Once an account dropped below 3 we submitted it for cancellation. Usually all three accounts on that domain were doing badly too. Because checks were five days apart, the failure day we recorded is approximate: an account could have slipped at any point after its previous check.
| Measurement | How we defined it |
|---|---|
| Monitoring tool | MailReach |
| How often | Every 5 days, for every account |
| Failure line | MailReach score below 3 |
| What we also saw | Reply rate close to 0% |
| What it meant | Poor deliverability and spam placement, not necessarily a suspension |
| What we did | Submitted the account for cancellation |
| Domain pattern | Usually all 3 accounts on the domain were doing badly |
| Lifespan | From the first day of outbound sending to the first check below 3 |
In February we launched three groups with 7, 14 and 30 days of warm-up. As soon as a group finished its warm-up, it started emailing cold prospects, with no head start on past responders.
Generally, the shorter the warm-up, the sooner the accounts stopped working. Some 7-day accounts crossed the failure line around day 15 or 16 of sending, and we don't have a single figure for that whole group. The 14-day group lasted about 28 days. The 30-day group averaged roughly 90 to 120 days, which helped justify the extra warm-up cost.
| Warm-up | How long the February accounts lasted |
|---|---|
| 7 days | Some failed around sending day 15 or 16 |
| 14 days | About 28 days |
| 30 days | An average of roughly 90 to 120 days |
The second test asked whether an account's first few days of real sending make a difference. Over several years we had kept a database of roughly 50,000 to 60,000 people who had responded positively to earlier campaigns. They were previously engaged contacts: not new cold leads, and not necessarily live sales opportunities.
We verified and cross-checked those records before reusing them, and our rule was never to email anyone more than once every 90 days. We didn't email the whole database at once either. About 3,000 to 4,000 of these contacts were approached per sending day, five days a week, across the sending operation.
Each March group spent only its first 2 or 3 sending days on these people, and about 4 to 5% replied. Then the very same accounts moved to ordinary cold outreach, aimed at a broad range of job titles and offering B2B services such as accounting, insurance, CRM and payroll. Replies after the switch were around 2.5%. We don't have a reply rate for the February groups to compare it with.
| Phase | Who got the emails | How long | Reply rate we saw |
|---|---|---|---|
| Start | People who had replied positively before | First 2 to 3 sending days | About 4 to 5% |
| After the switch | New cold prospects, broad job titles | The rest of the account's life | About 2.5% |
Our working theory is that those early replies may have helped the accounts build a better sending pattern. The test shows that starting on past responders went along with longer account life. On its own, it doesn't prove how Gmail judged the messages.
| Warm-up | How long the March accounts lasted |
|---|---|
| 7 days | 45 days or longer |
| 14 days | About 90 to 100 days |
| 30 days | More than 6 months |
Approximate, drawn from the tables. The lighter end of a bar is a range, an arrow means "or longer", and the short mark for the February 7-day group shows where some of its accounts failed.
| Warm-up | February: cold from day one | March: past responders first, then cold |
|---|---|---|
| 7 days | Some failed around sending day 15 or 16 | 45 days or longer |
| 14 days | About 28 days | About 90 to 100 days |
| 30 days | An average of roughly 90 to 120 days | More than 6 months |
Longer warm-up went with longer account life in both tests.
And every group that started on past responders outlasted the cold-first group with the same warm-up.
Only one of these figures is an average: the roughly 90 to 120 days for the February 30-day group. The others are the figures our team reported, and we won't call them averages until we've worked them out from account-level records.
A longer warm-up pushes back your first cold email and costs more up front. But an account that stays effective for months instead of weeks needs replacing less often. The February results suggested a 30-day warm-up was worth that cost.
The March results suggest it also matters who an account emails first. A few days on previously engaged people went with longer account life afterwards, and the cold outreach that followed still got about 2.5% replies.
Spasovska, E. (2026, October). How long 1,800 new Gmail accounts lasted in cold outreach. Artemidos. https://artemidos.com/research/email-warmup-study/