An abandoned cart email is a message sent to a shopper who left something behind without buying.
Every platform agrees on that. They disagree on which shopper.
One sends it when an item goes into the cart, then moves that shopper into a different sequence the moment checkout starts. Another sends it only once checkout has started. A third waits ten minutes after an email address is typed.
Three products, three different groups of people.
The abandoned cart email, as published on 2026-08-24. Omnisend splits the job into two automations with two triggers. Klaviyo calls one flow “abandoned cart” and recommends triggering it on the checkout event — which is Omnisend’s other automation. Shopify defines abandonment by an email address and a ten-minute clock. Mailchimp offers three fixed delays and one message. And the 70% statistic used to justify all of it counts carts, not people you are allowed to email.
What is an abandoned cart email?
It is an automated message triggered by a shopper who started buying and stopped.
Klaviyo’s help centre puts it plainly. The flow is “a message or sequence of messages sent to someone who added an item to their shopping cart, but failed to complete the purchase.”
Mailchimp describes its own the same way, as an email “to customers who add items to their cart but don’t check out.”
Both put the behaviour at the cart. One of them then fires somewhere else.
What counts as an abandoned cart?
Here they come apart.
Omnisend runs two separate automations and says who each one is for. The first “targets contacts who add items to their cart but don’t start checkout.”
The second “targets contacts who begin checkout but don’t complete the purchase.”
The first fires on “Added products to cart” and ends the moment checkout begins. Its exit condition is “Placed Order or Started Checkout”.
So starting checkout does not advance that shopper through the sequence. It removes them from it.
Which event does “abandoned cart” actually fire on?
On the checkout, usually — even where the flow is named for the cart.
Klaviyo: “The most common type of abandoned cart flow uses the” Started Checkout “event as a
trigger.” Most pre-built flows follow suit, though “some ecommerce integrations track the”
Added to Cart “metric, which triggers when someone adds an item to their cart.” The checkout
event “triggers when someone enters their information at checkout and moves to the next step of
the process”, which is later than the cart.
Read those two vendors together and the naming inverts. The flow one company calls abandoned cart is triggered by the event the other company calls abandoned checkout.
Same two words. Different shopper.
Why does one page define it two ways?
Because the definition and the mechanism were written for different moments.
The sentence defining the flow describes someone who “added an item to their shopping cart”. The recommended trigger fires later, once contact details have been entered and the shopper has moved on through checkout.
A shopper who fills a cart and closes the tab matches the definition. On the default setup, they never enter the flow.
How long before a cart counts as abandoned?
Only one platform here publishes a hard threshold.
Shopify’s is explicit: “A checkout is considered abandoned if it remains incomplete for more than ten minutes after the customer has provided their email information.”
That sentence carries two conditions. Ten minutes, and an email address already typed.
The others publish send delays instead — when the message goes out, not when the cart is considered lost.
What has to be true before the email can be sent?
The shopper has to be someone the platform already recognises.
Omnisend is the most direct about it. The automations “only work for identified contacts – those Omnisend can recognize”, and a contact becomes identified in four ways: clicking a link in one of its emails, signing up through one of its forms, entering an email at checkout, or placing a first order on WooCommerce.
Then the consequence: “Anonymous visitors who add items to the cart without any of these actions cannot receive abandoned cart emails.”
Klaviyo says the same from the other side. “Contact information is required for creating a profile.”
This constraint decides the size of the opportunity.
When should the first message go out?
Between two and four hours, on the only vendor here that publishes a window.
Klaviyo’s pre-built flows “include a 4-hour time delay before sending the first message”, and its written advice is to send “your first message between 2–4 hours after someone starts a checkout, and then sending a second email 20-48 hours later.”
Mailchimp does not offer that range. It offers three fixed options: “You can choose to send your abandoned cart email 1 hour, 6 hours or 24 hours after the customer leaves their cart.”
Why do the published delays disagree?
Because two of them are set by data plumbing, not by shopper psychology.
Klaviyo’s floor is a sync artifact: “Some integrations sync once every hour”, so it recommends “a time delay of at least 1 hour and 15 minutes to accommodate different sync timings.”
Omnisend gives the same kind of reason for a different number — add “a delay of at least 30 minutes before your first message to give the order event time to sync and trigger the exit condition.”
Neither floor is about how long a shopper needs. Each exists because the data behind it lags.
Is there a timing model that follows the shopper?
One, and it is keyed to browsing rather than the clock.
Omnisend’s “Trigger Inactivity Time” setting “defines how long a shopper must remain inactive in the store before Omnisend considers them to have finished shopping and triggers the automation.”
It exists so the email reflects the shopper’s “complete cart”, not just the first item added.
How many messages should the sequence have?
Two or three, on the only vendor here that calls its number a tested recommendation.
Klaviyo: “Based on research from thousands of Klaviyo ecommerce customers, we have found that sending 2–3 abandoned cart messages in your flow leads to optimal performance.” No sample size, method or underlying number appears on the page.
Omnisend’s checkout presets ship with three emails, or three plus an SMS. Mailchimp’s classic abandoned cart email is “a single follow-up email” unless you move to one of its other products.
What goes inside the message?
The thing the shopper left, printed back to them.
Klaviyo’s default flow “contains a dynamic content block that is unique to each ecommerce integration”, carrying the image, title, quantity and price of each item. Mailchimp is stricter: “You must include the Cart content block in your abandoned cart email”, and “Custom-coded templates are not compatible with the abandoned cart email.”
Underneath the personalisation, Shopify states that “Items in a customer’s cart at the time that they abandoned the checkout aren’t saved.”
Does the button go back to the cart?
Not always, and the default is where this bites.
Omnisend documents its own: the “Shop Now” button in the product block “links to the individual product page, not to the contact’s full cart.” Recovering the whole cart takes a second button, pointed at a separate variable.
Klaviyo’s return link “will depend on your integration”, and only “Some integrations allow you to rebuild a customer’s cart fully no matter which device they click the link from.”
A shopper who abandoned a full cart and lands on one product page has not been recovered.
Should the first email carry a discount?
The published advice says no, for a reason worth keeping.
Klaviyo: “we don’t encourage you to include a coupon in the first email. This can train customers to abandon carts in order to get the discount and is oftentimes unnecessary.” Hold it for the final message, or for people who have never bought.
Omnisend’s value-split preset selects differently — “contacts with a higher order value receive emails that include a discount”.
One picks by purchase history, the other by order value. They are answering different questions.
What stops the email after someone buys?
A filter you have to add yourself, and it does not always win the race.
Klaviyo requires it: “You need to add a filter to prevent people who complete a purchase from continuing into the flow.” The documentation says you will typically see “a large number of people who enter the flow being skipped”.
Both Klaviyo and Omnisend then document the gap. An order placed between the trigger and the send “may not cancel the other in time”.
Mailchimp states the opposite for its own queue. If the purchase lands before the send, “they’ll automatically be removed from the queue.”
What do the platforms admit they miss?
Several things, and each one is written down in their own documentation.
Omnisend does not track removals, so its emails “will display the products that were in the cart when the contact last added items or started checkout, even if they later removed some items.”
It also caps a contact to one abandonment automation per day, and does not fire retrospectively: shoppers who abandoned before you switched it on “will not receive emails.”
Shopify publishes six situations where no recovery email is sent at all — among them a shipping address it cannot serve, an empty inventory, and a shopper who gave a phone number instead of an email.
Is 70% of carts really abandoned?
The figure is real, published, and computed from a list you can read.
Baymard Institute publishes “70.22% – average documented online shopping cart abandonment rate”, described as “an average calculated based on 50 different studies containing statistics on ecommerce shopping cart abandonment.”
Those studies range from 55.00% to 84.27%. The list is headed 2026 and stamped “Last updated: September 22, 2025”; its newest input is one 2025 study, and its lowest value dates from 2010.
A single midpoint across that spread is the first thing a benchmark hides.
Published versions of that number disagree, and two of the three reconcile. Baymard’s own article says the rate “currently sits at 70.19%”, and a second Baymard page repeats it while calling the source list “A reference list of 49 cart abandonment statistics” — the list itself says 50. Average the 50 values it prints and you get 70.2192, which rounds to 70.22%; remove the one entry dated 2025, 71.72%, and the remaining 49 average 70.1886, which rounds to 70.19%. Same list, two moments. A third figure resists that explanation: the same article describes research done “to figure out why those 68.8% of shopping carts are abandoned”, and nothing beside it says what that was an average of. The copies drift too — Solidgate carries the rate correctly and prints a reasons table, sourced to the same institute, at different values, where nine of the ten shared rows differ and its prose puts window shopping at 43% against the live chart’s 42%. Nothing there is fabricated — the guide’s own 18% payment share adds up against the table it printed, and not against the live one.
Does the 70% describe your shoppers?
Only if your denominator matches theirs.
Two independent publishers print the same arithmetic. Solidgate: “Divide completed purchases by total carts created, subtract that figure from one, then multiply by 100. If 300 out of 1,000 carts result in a purchase, the abandonment rate is 70%.”
Dynamic Yield’s version is the same operation, worded differently: divide purchases by “the number of shopping carts created within the same period”, subtract from one, multiply by 100.
The denominator is carts created. Not sessions, not shoppers, and not subscribers.
Why can’t the email reach every abandoned cart?
Because the two populations are not the same.
Carts created includes every anonymous visitor. The email requires an address — Omnisend says anonymous shoppers “cannot receive abandoned cart emails”, Shopify requires that “the customer has provided their email information”, and Klaviyo needs contact information before a profile exists at all.
So the mailable group is a subset of the counted group. No platform publishes how much smaller.
The direction is not in doubt. A recovery plan sized against 70% is sized against the wrong number.
What do shoppers say they left for?
The most common answer is that they were not ready to buy.
Baymard’s survey of “1,083 US adults who have shopped online in the past 3 months” puts “I was browsing / not ready to buy” top at 42%, ahead of “Extra costs too high (shipping, etc.)” at 40%.
Then “Delivery was too slow” 20%, “I didn’t trust the site with my credit card” 19%, “Had to create an account” 18%, “Too long/complex checkout process” 17%.
The top answer is not a checkout problem. Baymard calls that segment “almost impossible to reduce through a better checkout design”.
What should you check before trusting a recovery benchmark?
Three things, in this order.
Which event it counted. A rate built on checkout starts and a rate built on add-to-cart are not the same measurement, whatever the label says.
Which population it counted. Carts created, or contacts you were permitted to email. No platform publishes the distance between the two.
When the number was read. The average shifted when the list grew, and the pages quoting it shifted at different times.
Where does that leave the flow?
Worth running, and worth sizing honestly.
Nothing above argues against sending it. The shopper had already chosen the product.
What the documentation does argue against is the arithmetic people build on top of it. Seventy per cent of carts are abandoned. Only the ones attached to an address can be emailed at all.
Pick the trigger deliberately, know which shopper your platform is actually reaching, and measure the flow against your own last quarter — the same discipline a lifetime-value figure needs, and for the same reason. A number only compares to itself.