Abandoned Cart Subject Lines: What to Write and Test

Abandoned cart subject lines get written the wrong way around on most teams: someone drafts three, one for each message in the flow, and only afterward notices that all three are trying to do the same job. This is a set of illustrative subject lines you can adapt, organised by where they sit in the sequence and who they’re going to, plus the setting and the sequencing mistake that causes most of that overlap. Every example here is written for this article, not lifted from a real brand’s send, and none is presented as a proven performer — each is a starting point to test against your own list, in the way abandoned cart email examples covers for full message structure and what an abandoned cart actually is covers for the trigger underneath it.

What’s the difference between a message-one, message-two and message-three subject line?

Each message in an abandoned cart sequence exists to make a different case, and the subject line should signal that difference before the email is even opened. Message one is a plain reminder — the shopper likely still remembers browsing, so the subject line’s job is recognition, not persuasion. Message two, sent once that recognition has worn off, needs an actual reason to come back: a real detail about the product or the cart, not a rephrased reminder. Message three, the last touch before the sequence ends, is where urgency or an incentive belongs, if the flow has one to offer — and where it doesn’t, the honest move is a plain final nudge rather than an invented deadline.

Illustrative examples, organised by position:

Message one — plain reminder, sent soon after abandonment

• “You left [item] in your cart”
• “Still thinking it over?”
• “Your cart’s saved, for now”

Message two — a specific reason to return, sent a day or so later

• “[Item] is almost out of stock”
• “Quick question about your size”
• “Here’s what other shoppers picked instead”

Message three — urgency or incentive, sent as the sequence closes

• “Last chance: 10% off your cart” (illustrative discount — set the real figure in your platform, not this line)
• “Your cart closes tonight”
• “One more nudge, then we’ll leave it be”

How do subject lines change by segment?

Segmenting the flow itself adds a layer on top of those three positions, because a first-time shopper, a repeat customer, and a subscription abandoner are each weighing a different decision. A repeat shopper already trusts the brand, so a subject line built around brand introduction wastes the one thing they don’t need reassurance on; a subject line acknowledging the relationship — “Welcome back to your cart” instead of “Still thinking it over?” — fits that segment better at message one. A first-time shopper is more likely to respond to a subject line that reduces risk — mentioning a returns policy or a guarantee, if the brand has one worth naming — since trust, not memory, is the open blocker for that group.

A subscription abandoner is deciding whether to commit to a recurring purchase, not just a single item, which changes what “urgency” even means. A message-three subject line built for a one-time cart — “Last chance: 10% off your cart” — undersells a decision that’s really about ongoing commitment; something closer to “Still deciding on your subscription?” matches what the shopper is actually weighing. A high-AOV cart, meanwhile, is often abandoned for a reason unrelated to interest — a shipping cost surprise, a need to check with someone else before a larger spend — so a message-two subject line for that segment does better naming a concern directly (“Questions about your order?”) than pushing urgency a shopper in a considered-purchase mindset is likely to ignore.

Where does preview text fit, and what should it say that the subject line doesn’t?

Preview text is the second line most inbox clients show next to the subject line, and its job is to extend the subject line, not restate it. A subject line reading “You left [item] in your cart” paired with preview text reading “Don’t forget your cart!” wastes the preview entirely — it repeats information the subject line already gave. The same subject line paired with preview text reading “Still available in your size, for now” adds a genuinely new piece of information the recipient didn’t have from the subject line alone, which is the actual test for whether a preview line is doing its job.

Most platforms, Klaviyo included, let a preview text field sit empty, in which case the client falls back to pulling the first visible text in the email body — often alt text, a hidden preheader block, or the first line of copy, none of which is written for this purpose. Set the preview text deliberately rather than leaving that fallback to chance, and check how it actually renders on both a phone-width inbox and a desktop one, since the two commonly truncate at different lengths.

What personalisation tokens are worth using, and where do they break?

The tokens available inside a Shopify-Klaviyo abandoned cart flow typically include first name, the abandoned product’s name, and sometimes a countdown or cart-value token, depending on the platform and the data captured at checkout. The token that breaks first, almost every time, is product name on a multi-item cart: a single-item cart resolves cleanly to “Your Product X is waiting,” but a three-item cart resolves to something like “Your Product X and 2 other items are waiting,” and if that fallback text was never previewed, it ships looking noticeably worse than the single-item version it was designed and tested against.

First name breaks the same way for guest checkouts with no name captured, or for a shopper whose name field held something unusable — a business name, a typo, an empty string that renders literally as blank in the subject line. The setting almost every team gets wrong here is proprietary to how these platforms are actually built: writing and approving the message-one, message-two and message-three subject lines by looking only at the single-item, named-shopper preview, then shipping without ever previewing the multi-item or no-name fallback case that a meaningful share of real abandoners will actually trigger. Check both fallback paths in Klaviyo’s dynamic tag preview before a subject line goes live, not after a shopper reports it looking wrong.

Emoji, punctuation and length — what actually needs testing versus what’s just style?

Whether an emoji helps, hurts, or does nothing to a subject line is not settled in general — it depends on the inbox providers your list actually uses and how the emoji is placed, so it belongs in the test queue rather than in either a house-style ban or a house-style habit. One relevant emoji placed at the end of an otherwise plain subject line is a different test than several stacked at the front, and treating those as the same experiment will produce a confusing result that doesn’t generalise to either case. The same logic applies to an exclamation point or all-caps word: test it as its own isolated variable rather than assuming it reads as urgency to every recipient the way it reads to the person who wrote it.

Length matters mostly because of truncation, not because a shorter line is inherently better. A phone-width inbox commonly cuts a subject line well before a desktop client does, so the words carrying the actual information — the product reference, the specific reason, the named incentive — belong before the point most inboxes are likely to cut, with any flourish or brand voice placed after it, where losing it costs nothing.

How do you A/B test abandoned cart subject lines properly?

Test one variable at a time. A test comparing “You left [item] in your cart 👀” against “Still thinking it over?” changes wording and emoji and sentence structure simultaneously, so a result either way tells you nothing about which change caused it. Isolate a single variable — wording, personalisation, emoji, length — hold everything else fixed as the control, and only then read the result as attributable to that one thing.

Let the platform’s significance calculation decide, not the first day’s numbers. Klaviyo’s built-in split test on a flow message tracks a statistical-significance threshold before it declares a winner, and reading an early lead as the answer — stopping the test after a day because one variant is ahead — is the fastest way to promote a losing variant by accident. A before/after comparison across two separate weeks or months is not a controlled test at all, since seasonality, traffic source, and promotional activity all move between those two windows independently of the subject line change.

Retire the loser and queue the next test rather than treating the winner as finished. The winning subject line becomes the new control, the queue’s next variable — often preview text, tested separately from subject line for the same reason wording and emoji shouldn’t be bundled — runs against it, and the bank grows by replacement rather than by endless addition. A subject line bank that only ever adds new lines without retiring old ones stops being a test programme and becomes an unmanaged pile of variants nobody is learning from.

Subject lines are the visible layer of a flow that has to be correct underneath them first — the trigger, the segment filters, the fallback text a token resolves to when the data behind it is incomplete. Getting that structure right for every checkout path a store actually has, not just the default single-item, named-shopper case a subject line reads cleanly for, is the kind of build we do as part of lifecycle flows.