Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Email Marketing › Email Automation

Building Abandoned Cart Flows

A step-by-step tutorial for abandoned cart and browse flows: identifying the shopper, triggers and timing, cart content, and incentive policy.
Tutorialintermediate12 min readUpdated

Before you start

  • A working welcome series. The welcome series tutorial establishes the pattern of trigger, spacing, exits and tests that this flow reuses, and the shop needs the sign-up and consent plumbing it relies on. The automation hub explains how the two flows share the list, the calendar and the priority rules between them.
  • Events from the shop. The sending platform needs to receive, as they happen, at least these: product viewed, item added to cart, checkout started, and order completed, each with the product identifiers, quantities and values involved. Most shop platforms send them through an integration; a custom shop sends them from its data layer or back end.
  • A way to tie events to an email address, covered in the first step. Without identity, there is no one to email.
  • A consent position. Reminders about an unfinished purchase are marketing in most readings, as the promotional versus transactional comparison sets out, so decide with whoever owns compliance which shoppers can be emailed and on what basis. Rules vary by jurisdiction.
  • A margin figure and an agreed incentive budget, because the incentive policy step needs to know what a discount costs.
  • The newsletter template and product image assets, so that the cart email can show what was left behind in the shop's own style.
  • Analytics that can attribute orders to email, so that the measure step has something to read; the email metrics and testing guide covers how.

Building the flow

  1. Identify the shopper

    A cart can only be followed up if the shop knows whose it is. Logged-in customers are identified by their account. Returning visitors can be recognised from a previous session where the same device clicked through from an email or logged in, provided the consent position allows that recognition and the privacy notice describes it. New guests are identified at the point they type an email address into the checkout, which is why capturing the address early in the checkout form, before payment, matters so much to the size of the flow. Do not treat a typed address as marketing consent on its own; the entry rule in the next step decides who can be emailed. Record for each identified shopper the basis on which they can be contacted, and make the flow read it before anything is sent.

  2. Set the cart and browse triggers

    The cart trigger fires when an item has been added and no order has followed within a defined wait. Define a cart precisely: at least one item with a value above a floor the business sets, so that a single low-value accessory does not start a three-email sequence. Define abandoned as the absence of an order event, not the absence of a checkout step, because many shoppers never reach checkout. The browse trigger is softer: a product page viewed more than once, or several products in one category viewed in a session, with no add-to-cart. Keep the two as separate flows with the cart flow taking priority, and apply the entry rules to both: identified, contactable on a recorded basis, not a recent purchaser of the same item, and not already in the flow for a different cart.

  3. Decide the timing

    The first cart email goes soon after abandonment, while the decision is still live: within the hour is the common choice, and the wait should be long enough that a shopper who is merely comparing delivery options is not interrupted. The second follows the next day, and the third, if there is one, a few days after that. Browse emails wait longer, usually until the next day, because the intent is weaker and an instant email about a page someone glanced at feels like surveillance. Set sending windows in local time so that neither flow sends overnight. Cap the series: a cart that has not converted after the third email is a decision made, and a fourth message spends goodwill for nothing. Write the timing into the flow's notes alongside the reason for each gap, so that later changes are deliberate.

  4. Write the email content

    The first email is a reminder, not a pitch: the items left behind, with images, names, variants and prices pulled from the cart, and a button that restores that exact cart. A short line that acknowledges the obvious is enough; shoppers know what happened. The second email resolves objections: delivery costs and times, the returns policy, sizing help, reviews of the items, a way to ask a question. The third makes the ask once more and, only if the policy from the next step allows it, carries the incentive. The browse email shows the product viewed and a few related items, with the lighter tone of a suggestion. Every email in both flows carries the full footer, the unsubscribe link and a preference option, because this is marketing email. Keep subject lines plain and specific to the item where the platform can merge it in.

  5. Set the incentive policy

    Discounts recover carts and train shoppers to abandon. A regular customer who learns that leaving a basket produces a code will leave every basket. Write a policy before building the third email, and build the flow to enforce it. Common positions: no incentive at all, relying on reminder and reassurance; an incentive only for first-time shoppers with no order history; an incentive only above a cart value where the margin can carry it; free delivery rather than money off, because it costs less and is harder to game; or a non-price incentive such as a gift with purchase. Whatever the policy, it reads from the customer's record, so that the same shopper does not see a code in the cart flow that the welcome series refused them. Review the policy against average order value (AOV) and margin every quarter, because the cost of a standing discount compounds.

  6. Set the exits on purchase

    The order-completed event exits every shopper from both flows, evaluated before each send and not only at entry. Check that the exit matches on the person, not on the cart identifier, because a shopper who abandons one cart and buys a different one has still bought. Add the other exits: an unsubscribe or complaint; the cart being emptied by the shopper; a reply to the email; and a stock-out of every item in the cart, which should either end the flow or switch the email to an alternative suggestion. Where the shop sends a transactional order confirmation, make sure it leaves on a different stream and that the cart email cannot arrive after it; the gap between an order and the next scheduled cart email is where most embarrassments happen, and the exit check before send is what closes it. A shopper whose item sold out can still be offered a request to hear when it returns, which the tutorial on back-in-stock and price-drop alerts builds as a separate, request-led flow.

  7. Set frequency and suppression rules

    A shopper who abandons carts twice a week is not two flows' worth of emails. Set a rule that a person can enter the cart flow only once in a defined period, with a new cart in that period updating the content of the email already scheduled rather than starting a new sequence. Set priorities between flows: cart beats browse, and both pause the welcome series. Cap total marketing messages per person per week across flows and campaigns, and decide which sends give way when the cap is reached. Suppress people with a recent complaint, a hard bounce or an open support ticket about an order. Suppress the browse flow for products the shopper has already bought. These rules are tedious to build and are the difference between a flow people tolerate and one they report.

  8. Test the flow end to end

    Use a test customer and a real cart. Add an item, leave, and confirm the first email arrives at the planned time with the correct items, images, variants and prices, and that the button restores the cart. Add a second item before the second email and confirm the content updated. Complete the order mid-sequence and confirm no further email arrives. Repeat for a guest who identified at checkout and for a browse-only session. Check every link's tracking parameters and that the unsubscribe works. Test the incentive branch with a first-time and a returning test customer. Once live with volume, the subject line and the second email's content are the usual candidates for an A/B Test, and the metrics guide covers how much volume a fair test needs; until then, run one version.

  9. Measure and review

    Define the flow's goal metric as the conversion rate (CVR) from entrants to orders within a window after the first email, and read it alongside click-through (CTR) per email, unsubscribes and complaints per email, the incentive cost against the revenue it accompanied, and the share of orders that came from shoppers who had already visited again before any email arrived. The last figure is the warning sign: a flow that reaches people on their way back anyway looks excellent and adds little. A holdout group that is identified and eligible but receives nothing is the honest measure, and the email metrics and testing guide explains how to size and read one. The conversion rate calculator turns the entrant and order counts into the rate. Put a review in the calendar monthly, and read the browse flow separately from the cart flow. Before the flow goes live, the email automation flow launch checklist includes the purchase exit, the life of any discount code and the holdout among its checks.

Flow template

The template below describes the cart flow for a shop with a no-incentive default and a first-order exception. The browse flow is a single email the next day, with the product viewed and related items, under the same exits and suppression. Paid remarketing on the same signals, where the shop also runs catalogue-based ads to people who viewed or carted products, is covered in the catalog ads guide and should share the purchase exit and the frequency logic described here, so that a shopper who bought is not chased by both channels. Which abandoners belong in the email flow and which are better reached by ads, and how to hold out a share of each, is set out in the comparison of abandoned cart emails and remarketing ads.

EmailTiming after abandonmentJobContentIncentiveExit check before send
OneWithin the hourRemindCart items with images and prices; restore-cart buttonNoneIdentified; contactable; no order; cart not empty
TwoNext dayReassureDelivery, returns, sizing, reviews, a way to askNoneNo order; items in stock; no reply
ThreeA few days laterAsk once moreCart items again; the strongest reason to buyOnly where the policy allowsNo order; not entered this flow recently; frequency cap respected
BrowseNext day after the sessionSuggestProduct viewed and related itemsNoneNo add-to-cart; no order; not already in the cart flow
Illustrative example — not a real result

A footwear shop's cart flow with a holdout

A footwear retailer launched a three-email cart flow with a first-order-only free-delivery incentive in the third email. Over 90 days, 24,600 identified shoppers abandoned a cart worth more than the £30 floor; 22,100 entered the flow and 2,500 were held out as a control that received nothing. Shoppers in the flow placed 2,430 orders within seven days of the first email, a rate of 11%, while the holdout placed 196, a rate of 7.8%. The difference, roughly 32 extra orders per thousand entrants, was the flow's real contribution, and at an average order of £84 it was worth about £59,000 over the period against an incentive cost of £4,100 in waived delivery. The second email drew more clicks than the first, which the team traced to the returns-policy paragraph, and the third email's unsubscribe count was double the others, which led to a rule that shoppers with two prior abandonments skip it.

To see where cart recovery sits in a wider sales plan, the email channel page for online sales sets it beside the welcome series, the promotional calendar and holdout tests.

When the incentive policy is in dispute, the abandoned cart discount test compares a plain reminder with a discounted one on margin per abandoner, with a no-email holdout as the reference.

If returning customers now wait for a code before buying anything, the diagnosis page for customers who only buy in the sale measures the discount habit by cohort and traces it to the cart flow and the store's other discounts.

To turn the flow's entrants, orders and held-out group into a rate and a real contribution, the conversion rate calculator set up for email flows runs both groups over one fixed window.

Verify it worked

  • Run a test customer through a real cart. The first email arrives at the planned time with the right items, images, variants and prices, and the button restores that cart.
  • Complete the order mid-sequence, then repeat after buying a different basket. No further email arrives either time.
  • Test the incentive branch. A first-time test customer sees the incentive only where the policy allows, and a returning test customer does not.
  • Check the report. Entrants, orders and a held-out group that received nothing all appear, so the flow's contribution can be read against the holdout.

If it goes wrong

  • A reminder arrives after the order confirmation. The exit was probably checked only at entry or matched on the cart. Evaluate the order-completed event before each send and match on the person.
  • Few shoppers enter the flow. The address is probably captured late, or the entry rule finds no recorded basis. Capture the address early in the checkout and check the contact record.
  • A shopper sees a code the welcome series refused them. The incentive policy lives in the template. Move it into the customer's record and make every flow read it.
  • The report looks excellent but the flow adds little. Without a holdout, it takes credit for shoppers who were coming back anyway. Hold out an eligible group that receives nothing and compare.

Next steps

Common mistakes

  • Emailing every typed address. An email in a checkout field is identity, not consent; the entry rule decides who can be contacted.
  • Treating the flow as transactional. The shopper has not transacted; the message is marketing and needs the footer, the unsubscribe link and a consent basis.
  • A discount in every third email. Regular customers learn to abandon; the policy belongs in the customer record, not in the template.
  • Exits checked only at entry. The reminder lands after the order confirmation.
  • Matching the exit on the cart, not the person. Someone who bought a different basket has still bought.
  • Instant browse emails. A message about a page someone glanced at a minute ago feels like being watched.
  • No frequency cap across flows. The welcome series, the cart flow and Thursday's campaign all arrive in one day.
  • Reading the flow's own report as its contribution. Without a holdout, the flow takes credit for people who were coming back anyway.
  • Never reviewing the browse flow separately. It behaves differently and is usually the one to trim.

Frequently asked questions

Do I need consent to send abandoned cart emails?

In most readings, yes, because a reminder about an unfinished purchase exists to prompt a sale and is treated as marketing rather than as a service message. An address typed into a checkout field is identity, not permission. Decide with whoever owns compliance which shoppers can be contacted and on what basis, record it against the customer, and make the flow's entry rule read it; rules vary by jurisdiction.

Should every abandoned cart email include a discount?

No. A discount in the standard sequence teaches regular customers that leaving a basket produces a code, and the cost compounds quietly. Write a policy first, such as no incentive, first-order only, or free delivery above a cart value, and build the flow to read it from the customer's record so that the same person is treated consistently across every flow.

How do I identify a shopper who has not logged in?

Through an address captured early in the checkout before payment, or through recognition of a device that previously clicked through from an email or logged in, where the privacy notice and consent position allow that. Capturing the address at the start of the checkout rather than the end is the single change that most enlarges the flow. Shoppers who remain anonymous cannot be emailed, and that is the correct outcome.

Terms used on this page

Tools to try

Frameworks

Where this page is used

Part of these learning paths

Also related

Related