Summarize this documentation using AI
Key Takeaways
- Send-time optimization is the practice of choosing when to send a message based on data rather than a fixed schedule, often per individual recipient.
- It is a real but modest lever. It improves engagement on messages you were already sending well, and it cannot rescue a message nobody wanted.
- Triggered messages largely make the question moot, because the trigger sets the timing and the moment of intent beats any statistical best hour.
- The bigger timing problem is frequency, not hour. Postscript recorded SMS campaign revenue per message falling from $0.31 to $0.14 across ten BFCM days while click rate held flat.
- Fix the flow library and the send frequency first. Send-time optimization is a refinement, not a strategy.
Send-time optimization is the practice of determining when to deliver a message for maximum engagement, using historical behaviour rather than a fixed schedule. Implementations range from a single best hour for a whole list, to per-segment timing, to a per-recipient model that predicts each individual's most responsive window.
It is a genuine capability and it is routinely oversold. Understanding where it fits saves a lot of wasted optimisation effort.
The three levels
- List-level. One best time for everyone, derived from aggregate opens and clicks. Crude, but better than a guess.
- Segment-level. Different times for different cohorts. Often the best return for the effort, since behaviour differs more between segments than within them.
- Individual-level. A model predicts each recipient's window and staggers delivery. Requires meaningful per-person history, so it works poorly on new subscribers, which is precisely where onboarding matters most.
Why it matters less than it sounds
Most of your revenue is triggered, and triggers set their own time
Across more than 183,000 Klaviyo accounts, automated flows produce nearly 41% of email revenue from 5.3% of sends (Klaviyo). A cart abandonment message should be sent relative to the abandonment, not at a statistically optimal Tuesday hour. Delaying a high-intent trigger to hit a better window trades a real advantage for a marginal one.
Send-time optimization therefore applies mostly to the campaign calendar, which is the smaller and lower-performing half of most programmes.
Open-rate signals got noisier
Privacy features that pre-fetch email content mean opens no longer cleanly indicate when a human engaged. Any model trained primarily on open timing is learning partly from machine behaviour. Click-based and conversion-based signals are more reliable inputs. Email deliverability covers the wider measurement consequences.
The timing problem that actually costs money
Frequency, not hour of day. Postscript's data across more than 17,000 Shopify stores recorded average SMS campaign revenue per message falling from $0.31 to $0.14 over the ten days from 25 November to 4 December 2025, while click rate stayed roughly flat at 3.2% down to 3.1% (Postscript).
Engagement barely moved. Revenue more than halved. People kept clicking and stopped buying. No amount of send-hour tuning recovers that, because the problem is saturation of the same audience with the same ask.
If you are optimising send time while sending five promotional messages a week to one segment, you are adjusting the seasoning on a meal that is already too large.
How to use it well
Order of operations
- Build the flow library, since triggered messages outperform scheduled ones regardless of hour.
- Fix frequency and segmentation, so fewer people get fewer, more relevant messages.
- Then optimise send time on the remaining campaign calendar.
Test it properly
Send-time optimization is exactly the kind of small effect that gets over-claimed by attribution. Run it as a randomised split and measure conversion rather than opens. A lift in opens with no lift in orders is not a result.
Respect time zones and quiet hours
The unglamorous version delivers most of the value: send in the recipient's local time, and never send SMS outside permitted hours. Getting time zones right beats any predictive model applied to the wrong clock.
What good looks like
A mature programme usually lands here: triggered flows fire on their trigger, campaigns go out in recipient local time within a sensible window, frequency is capped per person across channels, and send-time modelling is applied to the campaign calendar where there is enough history to support it.
That is a sensible use of the capability. It is also clearly a refinement layered on top of the things that actually move the number, which are covered in lifecycle marketing automation ROI in 2026 and triggered messaging.
The bottom line
Send-time optimization chooses when to send using data instead of habit. It is worth switching on where your platform offers it and worth testing honestly, and it will not fix a thin flow library or an over-mailed list.
Do the ordering properly: flows, then frequency, then hour. How to audit your lifecycle marketing program covers the sequence, and behavioral segmentation is usually the larger unlock sitting next to it.
Sources
- Klaviyo, 2026 Email Marketing Benchmarks by Industry (183,000+ customers)
- Postscript, 2026 SMS Benchmarks (17,000+ Shopify stores, 1 January to 15 December 2025)
Frequently Asked Questions
What is send-time optimization?
Send-time optimization is the practice of determining when to deliver a message for maximum engagement, using historical behaviour rather than a fixed schedule. Implementations range from a single best hour for a whole list, to per-segment timing, to a per-recipient model that predicts each individual's most responsive window. It is a genuine capability and it is routinely oversold.
Does send-time optimization actually work?
It is a real but modest lever. It improves engagement on messages you were already sending well, and it cannot rescue a message nobody wanted. It also applies mostly to the campaign calendar rather than to flows, which is the smaller and lower-performing half of most programmes: across more than 183,000 Klaviyo accounts, automated flows produce nearly 41% of email revenue from 5.3% of sends, and a triggered message should fire on its trigger rather than wait for a statistically better hour.
What are the levels of send-time optimization?
Three. List-level picks one best time for everyone from aggregate opens and clicks, which is crude but better than a guess. Segment-level sets different times for different cohorts and often gives the best return for the effort, since behaviour differs more between segments than within them. Individual-level models each recipient's window and staggers delivery, but it needs meaningful per-person history, so it performs poorly on new subscribers, which is exactly where onboarding matters most.
Is send time or send frequency more important?
Frequency, by a wide margin. Postscript's data across more than 17,000 Shopify stores recorded average SMS campaign revenue per message falling from $0.31 to $0.14 over the ten days from 25 November to 4 December 2025, while click rate stayed roughly flat at 3.2% down to 3.1%. Engagement barely moved and revenue more than halved. No amount of send-hour tuning recovers that, because the problem is saturation of the same audience with the same ask.
How should send-time optimization be tested?
As a randomised split measured on conversion, not on opens. It is exactly the kind of small effect that gets over-claimed by attribution, and a lift in opens with no lift in orders is not a result. Open signals have also got noisier, since privacy features that pre-fetch email content mean an open no longer cleanly indicates when a human engaged, so click-based and conversion-based inputs are more reliable.
