Best time to send marketing emails, and how to test yours

You have read three studies on the best time to send marketing emails and they gave you three answers. One says Tuesday morning, one says Friday, one says evenings win for clicks. They cannot all be right about your list, and in fact none of them is about your list. Each one is an average over millions of campaigns sent by other businesses to other people, and it describes those senders' subscribers, not yours.

That leaves two useful things a post like this can do. First, show you what the studies actually say, with their sample sizes and dates, so you can take a sensible starting time from them. Second, and this is the part every ranking page skips, show you how to test that starting time against your own list properly: which number to measure, what to change, how to split the list, and how many subscribers the test needs before the result means anything.

Why every study names a different best time to send marketing emails

Here are three sources that rank for this query, each read at its own page on 2026-09-24, and what each one claims.

Three things explain the spread, and they matter more than any single row.

The metric differs. A study built on open rates and a study built on click rates will name different hours even on the same data: MailerLite's own summary says opens cluster in the morning and clicks in the evening on most weekdays. If you sell something, the click is the number that pays, so a morning "best time" chosen for opens can be the wrong time for you.

The lists differ. A provider's average pools newsletters from creators, order confirmations from shops and B2B outreach into one number. Your subscribers are one kind of person on one kind of schedule. A study of two million campaigns tells you what a typical campaign did; it cannot tell you what your Tuesday will do.

The clocks differ. Some studies report the recipient's local time, some the sender's, and Twilio's holiday numbers are in Mountain time. A 9 AM peak for a US list is a 2 PM send for a UK subscriber and a midnight one in Sydney.

So treat a study as a place to start, not a verdict. The next section gives you a start; the rest of the post is the test.

The best time by day, from one study, and how little the days differ

The day-of-week completions people type ("best time to send marketing emails on friday", "on a monday", and so on for every day) get answered on the ranking pages with a different day each. Here is the one study above that publishes an hour and a figure for each day, from MailerLite's 2024 to 2025 data, so you can see the whole week at once.

Read the third column again. The gap between the best weekday and the worst is 1.2 percentage points, and the study's own note on the week says overall open rates hover between 46.5% and 47.5% on every day. If your list behaves anything like this one, the day you pick is worth about one open per hundred subscribers. The hour matters more than the day, the metric matters more than the hour, and your own list matters more than all three.

One more caveat on those open rates: they are high partly because of how opens are counted now, which is the next problem.

Test on clicks, because opens are no longer a clean signal

Since Apple's Mail Privacy Protection arrived in September 2021, Apple Mail can fetch a message's content, including the tracking pixel that records an "open", before the subscriber has looked at it. An open from an Apple Mail reader may mean the mail was read at 7 AM, or that a phone on a charger fetched it overnight. Pinpointe's page quotes a Litmus figure that the feature now affects roughly 55 to 60 percent of all opens, with the pixel firing through Apple's proxy servers whether or not anyone looked. MailerLite's study carries the same warning in its own words: since the feature's release on 20 September 2021 it recommends testing by click rate rather than open rate.

That decides the metric for you. For a send-time test, measure the action you actually want: clicks for most campaigns, replies for a plain-text B2B email, orders for a promotional email where your tool can attribute them. Opens are still worth a glance as a sanity check (a send that got no opens at all had a delivery problem, not a timing problem), but they are not what you compare.

How to test your own best send time, step by step

The ranking pages agree on "test it yourself" and stop there. Here is the test, in the order that avoids the mistakes that make results meaningless.

Do them in this order; a test that skips the random split or the waiting period measures something other than timing.

1. Choose the metric before you send. Click rate for most campaigns, or the conversion your tool can tie to the email. Write it down, so you cannot pick the flattering number afterwards.

2. Change one thing. Day or hour, not both. If you send Tuesday 10 AM against Thursday 8 PM and Thursday wins, you have learned nothing about whether it was the day or the hour. Test the hour first (it moves the numbers more in the study above), then the day.

3. Pick two candidates far apart. A study's morning peak against its evening peak (10 AM against 8 PM) is a real test. 10 AM against 11 AM is not: the difference will be smaller than the noise, and the arithmetic below says why.

4. Split the list at random. Not by signup date, not alphabetically, not "the old segment and the new one": those splits carry their own differences in engagement. Every mainstream email tool has an A/B split on a campaign; use it, at 50/50, with the identical email and subject line in both halves.

5. Wait before you read the result. An evening send has not had its morning yet when you check at breakfast. Read both halves at the same age. We suggest 72 hours after the later send as a rule of thumb; no study we read sets a waiting period, so treat it as our recommendation, not a measured figure.

6. Repeat, then log it. One send is one sample. Run the same pair on three or four campaigns, record each result in the log below, and only then move the winner into your default schedule. Then run the winner against the next challenger.

The log is the part that keeps you honest over months. Copy this table into a spreadsheet; the two rows are an example of the shape, not real results.

How big a list the test needs, with the arithmetic

This is the question none of the ranking pages answers with a number, and it is the reason most send-time tests "prove" whatever the sender hoped. A difference of half a percentage point in click rate, which is a real difference over a year, is invisible in one send to a small list.

The standard two-proportion sample-size formula, at 95% confidence and 80% power (the conventional settings for an A/B test), gives these figures. They are our arithmetic, not a study's, and they assume the baseline click rates shown.

The smaller the lift you want to detect, the larger each half must be; halving the lift roughly quadruples the list.

Now put a real list through it. Say you have 5,000 subscribers and a 2% click rate. A 50/50 split gives 2,500 per half, and 2% of 2,500 is 50 clicks. If the evening send is genuinely better at 2.5%, you would expect about 62 clicks. That gap of a dozen clicks is well inside the random swing between two sends of the same email, and the table says you need about 13,800 per half, 27,600 in all and five and a half times your whole list, to see it in a single send.

Three ways out, in the order to try them:

• Pool the sends. Four campaigns to the same split add up to 10,000 per half. That still does not reach 13,800 for the half-point lift, but it does reach the 3,800 needed to catch a lift from 2% to 3%, which is the size of difference worth changing a schedule for.

• Test bigger differences. Morning against evening, weekday against weekend. A 5,000-subscriber list can settle "2% against 4%" in about one send. It cannot settle "10 AM against 11 AM" in a year, so do not try.

• Accept a rougher answer. If your list is under a thousand, a formal test is not available to you. Send at the study's morning peak in your subscribers' local time, watch the click rate over a quarter, and change one thing at a time when it drifts. That is a slower test, not a worse decision.

What the arithmetic does not do: it does not tell you the lift is real for the next campaign, only that it was unlikely to be chance in the ones you ran. A sale email and a newsletter can have different best hours, so keep the log per campaign type once you have the volume.

Time zones, audiences, and the sends you should not test

If your list spans time zones, "10 AM" means nothing until you say whose. Most email tools can deliver by the recipient's time zone, so a 10 AM send lands at 10 AM in London and 10 AM in Chicago; use it, or your test result is really a test of where your subscribers live. A single-country list can ignore this.

B2B lists lean toward working hours because the mailbox is a work one; a consumer list that shops in the evening will show the opposite, which is exactly what the split between morning opens and evening clicks in the MailerLite data suggests. Neither is a rule. It is a reason to expect the two kinds of list to give different answers to the same test.

Leave these sends out of the test entirely:

• Triggered and transactional mail: a welcome email, an order confirmation, a password reset. Their best time is immediately, and a delay to hit a "good hour" costs more than it gains.

• Peak-season promotions: Black Friday, the week before Christmas, a launch day. Volume across every inbox is abnormal, so the result will not repeat in March.

• Sends to a re-engagement or cold segment: the click rate is too low for any split to reach the numbers above.

Several email tools offer a per-subscriber send-time feature that picks an hour from each person's past behaviour. It can help on a large list, and it is worth switching on once you have a baseline to compare it with. It is not a substitute for the test above: Mailchimp describes its version as built on sends through its platform plus data gathered from your subscribers, and MailerLite's as learning from historical data, so neither is testing the metric you chose against a control.

Check the mail arrives before you argue about when

A send-time test assumes both halves reach the inbox. If part of your list is landing in spam, the "losing" time may simply be the one whose mail was filtered, and no amount of scheduling fixes that. Run the four checks in why your emails are going to spam first: authentication, list quality, complaint rate, then content. A list that passes them gives you a test worth reading.

Then look at whether your tool can run the test at all. The email marketing software listing compares products on a fixed checklist, and four of its rows are the ones send-time work depends on: automated email sequences (so triggered mail goes out on its own clock), inbox preview and spam testing (so the two halves are not failing for a content reason), double opt-in (a cleaner list gives cleaner numbers), and a custom sending domain (the authentication why your emails are going to spam walks through). A tool that has those and a campaign A/B split is enough to find your own best time to send marketing emails, which is the only one that was ever going to be right.