Email marketing for ecommerce: the flows that matter and how to tell if they pay

The few email flows triggered by customer behaviour, how they differ from campaigns, and how to check that the revenue your email tool claims is real.

Growth economics
3 November 20266 min read

Email guides list a dozen flows and quote revenue shares from email tools. We cover the few that depend on customer behaviour: welcome, abandoned checkout, post-purchase, replenishment and win-back. This piece sits under our wider view of customer retention strategy for ecommerce.

What is the difference between flows and campaigns?

A campaign is a one-off send to a list or a segment. Klaviyo's help centre, which is a vendor's documentation, describes a campaign that way1. It also describes campaigns as proactive and flows as reactive and specific2. An email platform vendor's guide describes automation as emails sent when a customer does something3.

In plain English:

  • Flow. Starts when a customer does something, such as signing up, starting checkout or placing an order. It runs without you pressing send.
  • Campaign. Starts when you press send. A launch, a sale, a newsletter.

Klaviyo's getting-started guide lists welcome series, abandoned cart, post-purchase and win-back among the flows to set live first4. We start with the same ones and add replenishment where the product is consumable and the order data shows a reorder rhythm.

You need both. Flows catch the moment a customer shows intent. Campaigns tell people something that has just happened. Neither replaces the other.

Flows react to what customers do. Campaigns are one-off sends. You need both.

Welcome series: with or without a discount?

A welcome discount is a margin decision, not a copy decision.

Here is why. The discount comes out of the first order's contribution margin. Some subscribers would have bought at full price. For them, the discount is a gift. Some would not have bought at all without it. For them, the discount bought an order. You do not know the split until you measure it. That is a question of incrementality: how many orders exist only because of the offer.

Example (illustrative). A first order of €60 leaves €24 of contribution margin before marketing. A 10% welcome discount takes €6 of that. If half of the discounted orders would have happened at full price, you gave away €6 on each of those for nothing. Whether the other half is worth it depends on whether they come back.

Our view: test it. Send one group the series with the discount and another without it. Judge on completed orders, margin after returns and second orders, not on clicks or the first week's revenue.

For ZFD: our convention is to measure contribution margin before marketing spend, so any discount is read against the margin it eats.

How should an abandoned checkout flow work?

Klaviyo's training material describes the abandoned cart flow as starting when a checkout is started but not completed5.

In our view it is also the flow most likely to flatter itself. Some of those shoppers were coming back anyway. They got distracted, not talked out of buying. If your tool credits the order to the email, it is counting a sale that may have happened without it.

Practical points:

  • Do not default to a discount. Start with a reminder. Add an offer only if the holdout shows it earns its margin.
  • Match the product. Show what they left behind.
  • Stop the flow when they buy. Sending a reminder after a purchase is the quickest way to annoy a customer.

Post-purchase: reviews, cross-sell, the second order

Klaviyo's guide describes a post-purchase flow that starts when an order is placed and suggests splitting new from returning customers6. We agree with the split. A first-time buyer and a tenth-time buyer need different emails.

What the post-purchase flow can do:

  • Confirm and set expectations. What arrives, when.
  • Ask for a review. A review-app vendor's guide proposes a delay for review requests after delivery7. Time yours from the delivery date, not the order date. A customer cannot review what has not arrived.
  • Suggest what goes with it. Cross-sell only what fits the product bought.
  • Work towards the second order. In our view, that is where payback usually comes from.

Vendor guides do not agree on timing: an email platform's guide counts the review request from the purchase3, a review app's guide from delivery7. Test your own delays and read the result on orders and margin.

Replenishment and win-back

Replenishment. A replenishment email reminds a customer to reorder a consumable product. Time it from how often your customers actually reorder, not from what the product page says. A vendor's guide describes sending it ahead of the point the customer is likely to run out8. Stop it once they have bought again.

Win-back. A win-back campaign is a short series of messages to customers who have stopped buying. One vendor's guide says what counts as lapsed depends on the product category, and that contacts who do not respond should go to a suppression list because mailing them hurts deliverability9. We leave out the guide's day counts. They are one vendor's numbers.

Our view: the end of a win-back series is a suppression decision. Stop sending to contacts who did not respond. More on deliverability below.

Why is the email tool's revenue number too generous?

Because the tool counts its own sends. It cannot see what would have happened without them.

Ask your tool one question: what counts as email revenue in this report? An open? A click? Within how many days? Whatever the answer, a customer who was already on the way to buy can be credited to an email they barely glanced at.

Opens are shakier still. Klaviyo's help centre says Apple Mail Privacy Protection (MPP) loads the tracking pixel in advance, which inflates flow open rates10. If credit depends on an open, some of those opens may be Apple's, not your customer's.

The fix is a holdout. Keep a random slice of eligible customers out of a flow for a set period. Compare orders per customer and margin between the two groups. The difference is what the flow added.

Example (illustrative). In the sent group, 30 customers per 1,000 order. In the held-out group, 24 per 1,000 order. The flow added six orders per 1,000, not 30. The tool would credit the full 30.

You do not need to hold out every flow at once. Start with the one that takes the largest share of credit. For a wider look at the method, see incrementality.

Sources

  1. How to create and send an email campaign - Klaviyo Help CenterVendor
  2. Getting started with omnichannel campaigns - Klaviyo Help CenterVendor
  3. Features (Omnisend: email automation examples)Vendor
  4. Getting started with flows (Klaviyo Help Center)Vendor
  5. Move shoppers from consideration to purchase - Klaviyo AcademyVendor
  6. How to create a post-purchase flow - Klaviyo Help CenterVendor
  7. Post-purchase email strategy - Wiser ReviewVendor
  8. 7 replenishment email examples - OmnisendVendor
  9. Winback Email Campaign Strategy - LTV AIVendor
  10. Troubleshooting low flow open rates - Klaviyo Help CenterVendor

Key takeaways

  1. Flows react to what customers do. Campaigns are one-off sends. You need both.
  2. A welcome discount is a margin decision. Test it.
  3. Post-purchase flows should split new and returning customers.
  4. Email tools credit themselves generously. Check with a holdout.
  5. Consent and deliverability come before copy.

Next step

Want to know what your flows really earn? Start with a free audit of your retention set-up, or read how we work as an ecommerce retention agency. To discuss your flows, get in touch.