Smart Sending feels like a frequency cap, so most Klaviyo accounts treat it as one. It isn't. It's a per-message deduplication window that decides whether a single send skips a profile, and it has no view of how many messages that same profile already received this week from your other flows and campaigns.
The gap matters because over-mailing surfaces first in your complaint rate and your unsubscribe rate, and by the time those numbers move your inbox placement has already slipped. If Smart Sending is the only thing standing between your subscribers and message fatigue, you do not actually have a ceiling.
Key takeaways
- Smart Sending's default window is 16 hours for email and 24 hours for SMS and push, and each channel is governed separately.
- It runs on by default for campaigns but off by default on most flow messages, so several flows can stack on one profile in a single morning.
- A message skipped by Smart Sending is dropped, not rescheduled, so the feature can quietly remove a send you expected to land.
- A real frequency cap counts total messages per profile across every flow and campaign over a rolling period. Klaviyo has no native account-wide setting for that, so operators build it with segments and exclusions.
- Google asks bulk senders to keep the Postmaster spam rate under 0.10% and never reach 0.30%, and raw message volume is the fastest way to breach it.
What does Smart Sending actually do?
Smart Sending checks one thing at send time: did this profile already receive a marketing message on this channel inside the window. If yes, it skips them for that send. According to Klaviyo's Smart Sending documentation, the default window is 16 hours for email and 24 hours for SMS and push, and the three channels are configured independently under Settings.
Two mechanics catch operators out. First, the toggle lives per message, not per account. Campaigns ship with Smart Sending on, while individual flow messages default to off unless the pre-built flow set them otherwise, a point Klaviyo's flow frequency guidance makes explicit when it describes the "Skip recently emailed profiles" setting as an opt-in on each step.
Second, the skip is a drop, not a delay. Klaviyo does not automatically reschedule a message that Smart Sending removes, so a profile who trips the window simply does not get that email. That is fine for a duplicate blast and costly for a triggered reminder you assumed would send.
Why is Smart Sending not a frequency cap?
Because it never counts. A frequency cap tracks a running total against a budget, something like no more than three emails in seven days. Smart Sending only asks a yes or no question about the last few hours, then resets.
Walk the arithmetic. A profile receives an email at hour 0, another at hour 17, and a third at hour 34. Every one of those clears the 16-hour email window, so Smart Sending waves all three through inside roughly a day and a half. The subscriber sees three emails in two days; the feature designed to protect them saw nothing wrong, because it was never measuring the total.
Scope compounds the problem. Since flows default to off, a welcome step, a browse-abandonment step and a scheduled campaign can all fire on the same profile the same morning, each blind to the others. Smart Sending deduplicates within a channel and a window, not across your program.
Where does over-mailing actually cost you?
It costs you in the two metrics mailbox providers watch most closely: spam complaints and disengagement. Google's bulk sender guidelines ask senders to keep the spam rate reported in Postmaster Tools below 0.10% and to avoid ever reaching 0.30%, and complaint rate climbs with volume faster than almost any other lever you control.
Unsubscribes and quiet disengagement follow the same curve. A subscriber who gets three emails in two days does not reason about your windows; they hit unsubscribe or stop opening, and a shrinking engaged base drags your reputation down even when no one formally complains.
This is usually a specific offender, not a diffuse problem. If your complaint rate is drifting up, the discipline of finding which segment is burning your domain reputation matters more than adjusting Smart Sending, because Smart Sending was never the thing capping that segment's volume in the first place.
How do you build a real frequency cap in Klaviyo?
You build it with segments and exclusions, because there is no single account-wide "max N per week" switch. Three approaches carry most of the load, and they are not mutually exclusive.
The first is subscriber-declared frequency. You add a frequency property to your preference page, then honor it on send. Klaviyo's frequency segmentation guide walks through adding an "Email frequency" profile property as a radio block, then building Daily, Weekly and Monthly segments to include or exclude per campaign. Note what that implies: enforcement is manual, it covers campaigns rather than flows, and you have to add the exclusion to every send. You also need a default segment for profiles who never set a preference.
The second is a rolling suppression segment that does the counting Smart Sending will not. Define a segment such as "received an email at least three times in the last 7 days" and exclude it from campaign sends. This finally puts a real number on volume, though it still lives outside your flows unless you replicate the exclusion as a flow filter.
The third is discipline about the toggle itself: treat Smart Sending as a floor that prevents accidental same-hour duplicates, and treat a counting segment as the ceiling that governs weekly load. One stops the obvious mistake; the other governs the real exposure.
How do the frequency controls compare?
The four tools operators reach for solve different problems, and the common failure is assuming the first one does the last one's job.
| Control | What it does | Counts total volume? | Covers flows? | Main limitation |
|---|---|---|---|---|
| Smart Sending | Skips a send if the profile was messaged inside a 16h/24h window | No | Only where toggled on | Resets every window; blind across flows and campaigns |
| Frequency preference segments | Honors a subscriber-chosen cadence on the preference page | No, it honors a stated preference | No, campaigns only | Manual exclusion on every send; needs a default for unset profiles |
| Rolling suppression segment | Excludes profiles who crossed a message count in a rolling window | Yes | Only if replicated as a flow filter | You maintain the segment definition and remember to apply it |
| Declared cadence plus counting segment | Combines a stated preference with a hard volume ceiling | Yes | With flow filters, yes | Most setup effort; requires governance to stay consistent |
What breaks when you turn Smart Sending off on a flow?
You remove the last automatic guard against a same-day collision. Operators often disable Smart Sending on a high-value flow step, an abandoned-cart reminder for instance, to guarantee it lands. That guarantees delivery and also guarantees the step ignores every other message the profile received that day.
The visible symptom is duplicate or near-duplicate sends within hours. If a subscriber gets a campaign at 9am and an abandonment reminder at 11am with Smart Sending off on the flow, both land, and neither knew about the other. When that pattern shows up across an account, an audit of Klaviyo flow double-sending will usually trace it to a mix of disabled Smart Sending and overlapping triggers rather than a single broken flow.
The fix is rarely to toggle Smart Sending back on everywhere, because that reintroduces silent drops on messages you need delivered. It is to put a counting exclusion in front of the campaign side so the flow can stay authoritative for its trigger without competing blind.
What should you watch after you change any of this?
Watch complaint rate and unsubscribe rate by send, not just the account average, because a single over-mailed segment can sit under the account line while quietly poisoning your reputation. Google's 0.10% target is the number to hold against, and it moves before opens and clicks do.
Watch for skipped-send surprises whenever Smart Sending is on for a triggered message. If a welcome or reminder step shows lower send volume than entries into the flow, the window is eating messages you wanted delivered, and the answer is a targeted exclusion rather than a blanket window.
The trade-off to hold in mind is simple. Smart Sending buys you cheap protection against the dumbest mistake, a duplicate inside a few hours, at the price of silent drops. A counting segment buys you a genuine ceiling at the price of setup and governance. Serious programs run both and stop pretending the window is a policy.