Choose an email send time by testing when your audience responds to the campaign’s goal. For an ecommerce promotion, that means deciding whether you want more visits or more orders before declaring a schedule successful.
When should you send marketing emails? Start with a workable schedule for your audience, then test an alternative against your goal. Specify the time zone and offer deadline. For flexible campaigns, consider testing AI delivery against a fixed schedule. More recorded opens alone do not establish more sales, and a predicted hour is not a guaranteed winner.
Is there a best day and hour for every email?
Treat the best send time as a question about your audience and campaign, rather than a universal weekday or hour. Our recommendation is to compare a workable starting schedule with an alternative under the same conditions, using a result you chose in advance.
That turns “Should I send in the morning or evening?” into a decision you can investigate. For example, 10 a.m. and 7 p.m. are illustrative candidate times, not industry benchmarks. You might compare them if both suit the offer and the audience’s time zone. Neither earns a place on your calendar merely because it appears in this example.
Use the same approach for weekdays versus weekends. Name the alternatives and keep the campaign comparable. Comparing a discounted weekend sale with a full-price weekday newsletter would change the offer as well as the schedule.
If you have little usable history, choose a time your team can support and that leaves recipients room to act before the offer closes. Save the choice as a starting assumption. You can improve it with evidence later; you do not need an invented “best hour” to send the first campaign.
Should everyone receive the email at the same time?
Choose the scheduling method from the campaign’s constraints. A fixed local hour can be a useful candidate for an audience spread across time zones; personalized delivery is a candidate when the message can arrive over a window. An expiring offer needs an explicit deadline and reference clock.
These are different scheduling decisions:
| Method | What you are choosing | Check before scheduling |
|---|---|---|
| Fixed account time | One reference-clock time for the campaign | What that becomes in each audience region |
| Fixed recipient-local time | The same local hour for recipients in different locations | How the platform handles unknown time zones |
| Personalized delivery | Different predicted hours within an allowed window | What the model targets, feature eligibility and latest possible delivery |
Check which scheduling options your plan provides. Mailchimp’s Send Time Optimization illustrates why the target matters: it describes finding when recipients are most likely to open within 24 hours of the date you select. The feature requires Standard or higher and enough sent-email data, and is unavailable for automated emails. That is a documented scheduling feature, not evidence that its chosen hour will produce the most orders.
Klaviyo’s Personalized Send Time chooses recipient-local hours within a window, unlike its audience-wide Smart Send Time. It requires a Marketing Analytics or Advanced KDP subscription beyond the standard marketing application. As of October 2026, it applies to campaigns and is not available yet in flows; scheduling requires a future date starting tomorrow. Its documentation cautions against expiring offers; where no valid time remains on the scheduled date, delivery moves to the next day inside the configured window.
Before turning on an optimizer, write down the question it is supposed to answer. “Does this delivery window outperform our fixed schedule on orders?” is more useful than “Did the AI find the best time?” For the broader uses of prediction and generation, see what AI email marketing covers.
Which result should decide the winning time?
Choose the result that matches the campaign’s purpose before testing. For a sales promotion, our recommendation is a defined order-rate comparison where available, with clicks and unsubscribes as additional checks. A larger recorded open rate alone does not answer whether the schedule produced more sales.
Apple’s Mail Privacy Protection guidance says enabled protection prevents senders from seeing whether the recipient opened a message. Keep that limitation in mind when interpreting an open-time pattern.
Here is an illustrative example with invented counts. Each schedule has 2,000 assigned recipients. All rates below use that assigned-recipient denominator, rather than the platform’s possibly different reporting denominator.
| Result | Morning | Evening |
|---|---|---|
| Assigned recipients | 2,000 | 2,000 |
| Recorded opens | 800 / 2,000 = 40% | 700 / 2,000 = 35% |
| Unique clickers | 80 / 2,000 = 4% | 100 / 2,000 = 5% |
| Purchasers | 20 / 2,000 = 1% | 30 / 2,000 = 1.5% |
Morning leads on recorded opens; evening leads on purchasers. The purchase-rate difference is 1.5% − 1% = 0.5 percentage points, or (1.5 − 1) ÷ 1 = 50% relative difference.
Those calculations describe the invented results. They do not establish statistical significance, a reliable timing effect or improved long-term retention. The example shows why you should choose the outcome first: “50% higher” sounds decisive while concealing both the starting rate and the actual counts.
Keep the denominator and order-counting rule visible in your own comparison. Do not mix a rate based on delivered emails with one based on assigned recipients and call the difference a scheduling win.
How do you test a send time for your own audience?
Compare schedules within the same campaign, randomly assigning eligible recipients and keeping the subject, content and offer unchanged. Our suggested method is to define a region or time-zone group first, then compare timing within it so the reference clock stays clear.
This applies the NIST experimental-design principle of controlling known nuisance factors and randomizing others. NIST provides general experiment guidance, not an email-hour recommendation.
Use this experiment card before scheduling:
- Audience and clock: the eligible audience, region and time zone used for each candidate hour.
- Alternatives: your current schedule and one alternative; keep the subject, content and offer identical.
- Assignment: random allocation within the defined audience; record other campaigns or flows that could overlap.
- Outcome and follow-up: the primary result, its denominator, the order-counting rule and equal observation time after each group’s delivery.
- Decision: the smallest difference worth changing the schedule for; keep uncertainty visible rather than adopting whichever rate is numerically higher.
For a documented setup, Klaviyo’s campaign A/B guidance uses Campaigns → Create campaign → Email, then Create A/B test above the subject field and Test send times. Keep subject and content identical. Its placed order rate option requires a Placed Order metric and is unavailable for the personalized-variations strategy. Check the scheduling controls available for your test before assuming recipient-local timing is supported.
If you use the recipient-level feature described earlier, its control group is randomly selected and receives the same campaign on a non-personalized schedule you configure. It is a timing comparison, not a no-email group. Open the campaign’s Overview → View Impact tile; analytics appear at least 48 hours after completion, and reported open and click rates exclude Apple Mail Privacy Protection opens and automated bot clicks.
That reporting delay is a feature detail, not a universal experiment duration. Set your own follow-up rule before comparing outcomes, and give each group the same observation time after delivery. Save the audience, schedules and decision with the result so a future campaign does not inherit a conclusion stripped of its conditions.
When should you keep an explicit schedule?
Keep an explicit schedule when the message must arrive in time for a deadline. Treat event-triggered flow delays as a separate decision from a broadcast campaign’s clock time.
For an illustrative offer ending at 6 p.m., a promotional message scheduled for 7 p.m. cannot accurately present that offer as still available. Define which time zone the deadline uses and leave room for the recipient to act. Check delivery fallbacks as well as the intended window.
Likewise, do not transfer a campaign’s winning hour directly into a welcome flow or an abandoned-cart sequence. Decide the event-to-message delay separately, then test that question on its own terms.
Your next useful scheduling decision is small: choose one flexible campaign, name the reference clock, compare one alternative and retain the result with its conditions. If the audience, offer or deadline changes, reconsider whether that comparison still applies.