The ecommerce email flows that carry most of the revenue

Automated flows are 5.3% of a brand’s email sends and around 41% of its email revenue. That gap isn’t about copy quality. It’s that flows fire on intent and campaigns fire on a calendar. Rank flows by revenue per recipient instead of open rate and the usual ordering inverts, with the flow everybody opens sitting near the bottom.

Which email actually makes the money?

Flows, by a margin that surprises most people the first time they see it. Klaviyo publishes benchmarks drawn from more than 183,000 customers, and the split between automated flows and broadcast campaigns is the most useful table in the whole set.

Read the first two rows together. A twentieth of the sends produces two fifths of the revenue, which puts the per-send value of a flow at roughly twelve times that of a campaign. The revenue-per-recipient row measures the same thing directly and lands on eighteen.

Nothing here says campaigns are worthless. They reach an audience flows structurally cannot, because a flow only fires when someone does something and most of your list does nothing this week. What the table says is that effort spent on flows compounds and effort spent on campaigns is consumed on send.

Why does ranking flows by open rate get it backwards?

Because open rate measures whether an email was interesting and revenue per recipient measures whether it was timely, and timeliness is what a flow is for. The cleanest illustration is post-purchase, which posts the highest open rate of any flow and close to the worst revenue per recipient. The figure circulating for that revenue is around $0.41 to $0.47, which no primary publication appears to carry, so treat the exact number as unverified. The ranking itself — highest open, near-lowest revenue — is what the benchmark set states, and the ranking is the part that matters.

There’s a practical reason open rate has held on longer than it should. An open is recorded when a tracking pixel loads, and mail clients that prefetch images load it whether or not a human ever looked at the message. So the metric that is easiest to collect is also the one most contaminated by machines, while the metric that decides what to build next needs a revenue join nobody sets up by default.

What is an abandoned cart flow worth?

More per recipient than anything else most brands run. Klaviyo’s abandoned cart study, published 15 May 2024 across more than 143,000 flows, reports an average revenue per recipient of $3.65, a 3.33% conversion rate and a 50.5% open rate. The top decile earns $28.89 per recipient.

The vertical spread is wider still. Top-decile abandoned cart flows reach $75.66 per recipient in hardware and home improvement. Electronics sits close behind at $66.89. Those are high-consideration, high-ticket categories where a reminder arrives while the decision is still open, and they are the reason a single blended benchmark is close to useless for planning. If you sell $18 consumables, $75.66 is not a target you missed. It’s a different business.

Is email really a third of your revenue?

It depends entirely on the window, and the window is more generous than almost anyone quoting the number realises. Klaviyo announced on 7 August 2024 that its default attribution for email would become 5-day click and 5-day open. Not click alone. An email that someone opened and never clicked can be credited with a sale that happens up to five days later.

Two consequences follow. The first is that any flow reaching people who were going to buy anyway will look excellent, because those people open things. Post-purchase and order-confirmation-adjacent messages sit exactly there, which is one more reason their reported numbers need reading carefully.

The second is that the attribution is cooperative only inside Klaviyo. It doesn’t negotiate with Meta, Google or your analytics package, all of which are claiming the same orders on their own terms. Add the four dashboards together and you’ll comfortably exceed your actual revenue.

Nobody has published a defensible cross-brand figure for email’s true share of ecommerce revenue, and the ones circulating are vendor dashboards summed without adjustment. That’s worth knowing before you build a plan on one. There are several more widely repeated ecommerce figures in the same condition, collected in the numbers that do not survive checking.

Why is email an acquisition channel and not just a retention one?

Because 48% of flow revenue comes from new buyers, against 16% for campaigns, on Klaviyo’s benchmark population. Nearly half the money automated email produces is somebody’s first order.

That reframes where flow work sits in a budget. If flows are retention, they compete with loyalty programmes and win-back discounts for attention. If half their output is first orders, they compete with paid acquisition, and the comparison is brutal in email’s favour because the marginal cost of an additional send is a rounding error against a CPM.

It also changes what a subscriber is worth at the top of the funnel. An email address captured in September is a flow recipient through the whole of Q4, at a moment when paid inventory is at its most expensive. Nobody has published a September-versus-November comparison of cost per email acquired, which is a strange gap given how obvious the arithmetic is — the ramp in paid costs is well documented and what happens to CPMs from late October covers it.

What has to be true before any of this reaches an inbox?

Authentication, and a complaint rate under a hard ceiling. Google’s bulk sender requirements took effect on 1 February 2024 for anyone sending more than 5,000 messages a day to Gmail, and Microsoft introduced its own requirements for high-volume senders in May 2025. We could not verify Microsoft's exact enforcement behaviour against its own page, so treat that half as unconfirmed. The complaint ceiling below is Google's, and it is published.

The number to write on the wall is the spam rate. Google’s published threshold is below 0.30%, with 0.10% named as the target. Those are three complaints per thousand and one per thousand. A single badly segmented campaign to a cold list can clear 0.30% on its own, and the damage lands on the flows, which are the ones actually making money.

What order should you build them in?

1. 01

   Abandoned checkout, then abandoned cart

   Highest revenue per recipient of any standard flow at $3.65 on average, and the smallest audience, so it’s also the cheapest to get wrong. Three messages is the common shape. Send the first within an hour, while the tab is arguably still open.

2. 02

   Welcome

   Second because of reach rather than rate — everyone who joins the list passes through it, so it is the flow with the largest addressable audience you will ever build. It’s also where the 48% new-buyer share mostly comes from.

3. 03

   Browse abandonment

   Lower intent than a cart and a much larger trigger population, which is a trade you can only judge with your own numbers. Cap the frequency hard or it becomes the flow that generates your complaints.

4. 04

   Post-purchase

   Build it, but judge it on repeat purchase rate rather than on revenue per recipient. Held to the wrong metric it looks like a failure, and it’ll get cut by whoever reads the dashboard next quarter.

5. 05

   Win-back and sunset, together

   A win-back that never gives up is a complaint generator, and complaints are the one thing that can switch the whole programme off. Pair every win-back with the rule that suppresses the address when it ends.

The one thing to change this week

Pull revenue per recipient for every flow you run, put them in one column, and sort. It takes a few minutes and it usually produces at least one surprise, because the flow getting the most attention internally is generally the one with the best open rate rather than the best output.

Then check the attribution model the numbers were produced under before you act on the ordering. A five-day open window flatters the flows that reach engaged people, and the flows that reach engaged people are the ones you were already overrating. The same scepticism is worth carrying into every other benchmark you inherit, and a surprising number of them fall apart on inspection.