Nine flows: welcome, abandoned cart, post-purchase, browse abandonment, replenishment, win-back, price drop, post-review, and VIP or loyalty tier. Build them in that order rather than all at once, because the order runs from highest expected return per hour of build effort to lowest, and because your review capacity, not your ESP, is what caps how many you can ship. The reason most brands run three flows and not nine is not that they disagree about the other six. It is that nobody has counted what building them costs.
An automated campaign flow is a sequence of messages sent to one person in response to one trigger event, without a human scheduling each send. A trigger is the event that starts it: a signup, a cart with items and no order, a delivery confirmation, an inventory price change.
| # | Flow | Trigger | Channel | Purpose |
|---|---|---|---|---|
| 1 | Welcome | Email or SMS signup | Email + SMS | Convert a new subscriber to a first order |
| 2 | Abandoned cart | Cart created, no order within 1 hour | Email + SMS | Recover a session that already showed intent |
| 3 | Post-purchase | Order confirmed, then delivered | Set expectations, reduce support tickets, seed the second order | |
| 4 | Browse abandonment | Product page viewed 2+ times, no cart | Catch intent earlier than the cart | |
| 5 | Replenishment | Days since order = product consumption cycle | Email + SMS | Time the reorder instead of waiting for it |
| 6 | Win-back | No order in 2x the median purchase interval | Reactivate before the subscriber goes cold | |
| 7 | Price drop | Price change on a product in a wishlist or viewed set | Convert a price-sensitive segment on a real event | |
| 8 | Post-review | Review submitted | Route detractors to support, promoters to referral | |
| 9 | VIP or loyalty tier | Lifetime value or order count crosses a threshold | Email + SMS | Change the offer and the tone for your top decile |
Which three flows does almost every ecommerce brand already have?
Welcome, abandoned cart and post-purchase are the three flows that almost every ecommerce brand has already built, and all three are usually built once and never revisited.
They exist. That is different from working.
Welcome. The common failure is that it is a discount delivery mechanism with three reminders attached. One message with a code, two follow-ups saying the code expires. It teaches every new subscriber that the correct behavior is to wait for a discount, and it says nothing about why the brand exists, how the product is made, or which product the person should buy first. A welcome series that asks one question in message one and branches on the answer is a different asset from a welcome series that broadcasts the same code four times.
Abandoned cart. The common failure is timing collapsed into a single rule. One send at one hour, one at 24 hours, one at 48 hours, identical for a $19 accessory and a $600 sofa. Considered-purchase categories need a longer window and a different message: reassurance, returns policy, sizing, comparison. Impulse categories need the opposite. Running one timing rule across the whole catalog means the flow is wrong for most of it.
Post-purchase. The common failure is that it stops at the shipping confirmation, which the ESP is not even sending. Transactional messages come from the platform. The flow that matters starts after delivery: how to use the product, what to expect in week two, when to reorder, and the review request timed to the point where the customer actually has an opinion. Most brands send the review request on day three, before the product has been used.
Which six flows do most ecommerce brands never build?
Six flows are missing from most ecommerce programs: browse abandonment, replenishment, win-back, price drop, post-review and the VIP or loyalty tier.
None of them are technically hard. All of them require deciding on logic that nobody owns.
Browse abandonment. Trigger: a product page viewed two or more times in a session or across a short window, with no add to cart. The logic is a threshold decision. One view is noise. Two views on the same product in 48 hours is a signal. Worth building because it reaches people earlier than the cart flow does, and because the pool of browsers is far larger than the pool of cart abandoners. The risk is volume: without a suppression rule against the cart flow and a frequency cap, you will send this to the same person four times a week.
Replenishment. Trigger: days elapsed since the last order of a consumable, set to the product's actual consumption cycle. A 60-serving supplement is a 55-day trigger. A 200ml serum used twice daily is different. The logic requires a per-SKU or per-category cycle length, which is the work most teams avoid. Worth building because it is the only flow where the offer is a reminder rather than a discount, and the margin holds.
Win-back. Trigger: no order in roughly twice the median purchase interval for that customer's segment. Not a fixed 90 days. If your median repeat interval is 45 days, 90 is right. If it is 120 days, a 90-day win-back is annoying people who are behaving normally. The logic is a lapse definition, and it has to be computed from your own order data. Worth building because it is the cheapest flow to write and because the alternative is paying to reacquire the same person.
Price drop. Trigger: a price change on a product a person has in a wishlist, saved, or viewed within a defined window. The logic is a suppression rule as much as a trigger: exclude anyone who bought that product in the last 30 days, or you will send a discount notice to the person who just paid full price. Worth building because the event is real and dated, which makes the message legitimate rather than a manufactured urgency play.
Post-review. Trigger: a review submitted, branched on score. Detractors go to a human, not to an automation. Promoters go to a referral or user-generated content ask. The logic is the branch and the handoff to support. Worth building because it is the only flow that turns a service problem into a recovery rather than a public rating.
VIP or loyalty tier. Trigger: lifetime value or order count crossing a threshold you define. The logic is the threshold and what changes on the other side of it: early access, free shipping, a named contact, a different tone. Worth building because your top decile is already buying, and the flow is about retention economics rather than acquisition.
What order should you build ecommerce flows in?
Build in this order: welcome, abandoned cart, post-purchase, browse abandonment, win-back, replenishment, VIP tier, price drop, post-review.
The ordering principle is expected return divided by build effort, where build effort is measured in messages to produce and logic decisions to get signed off.
The first three come first because the triggers are unambiguous and the audiences are already in motion. Browse abandonment is fourth because it reuses the abandoned cart architecture almost entirely, so the marginal effort is low. Win-back is fifth because it is three messages and one lapse calculation. Replenishment is sixth because it needs per-SKU cycle data, which is a data project before it is a campaign project. VIP is seventh because it needs a threshold decision that usually involves finance. Price drop is eighth because it needs a product feed and a suppression rule. Post-review is last because it depends on a review platform being configured and on someone in support agreeing to take the detractor branch.
Adapt the order to your own constraints. If you sell one consumable SKU, replenishment moves to fourth. If you have no wishlist function, price drop drops off the list entirely. The reasoning is what transfers, not the sequence.
What does a nine-flow ecommerce program actually cost to produce?
A nine-flow program at four messages per flow is 36 messages, and once you multiply by channels, locales and supporting landing pages, a realistic build is over 100 individually produced and approved assets.
Here is the count for a brand running three locales.
| Component | Per locale | Across 3 locales |
|---|---|---|
| Email messages | 28 | 84 |
| SMS messages | 8 | 24 |
| Supporting landing pages | 6 | 18 |
| Total production assets | 42 | 126 |
Now the review arithmetic, which is the part that gets left out of every flow guide. A review round is one pass of an asset in front of its reviewers, ending in a decision. In-house, the reviewers are usually three: brand, legal or compliance, and a director who signs off. At two rounds per asset, 126 assets produce 252 review rounds and up to 756 individual decisions. At four minutes of attention per decision, that is roughly 50 hours of reviewer time, spread across people whose job is not reviewing your flows.
That is the constraint. Not the ESP. Cutting from three rounds to two on 126 assets removes 126 review rounds, which is a larger saving than anything you will get from switching platforms. The levers that actually move it are a brief specific enough that copy does not come back twice, a fixed reviewer list per asset type so nothing waits on someone who was never asked, and a QA pass before review so that rendering and personalization errors are caught by the producer rather than by the legal reviewer. A multichannel pre-send checklist handles the last one.
This is also where the plan and the production meet. Deciding which nine flows to run is campaign journey design. Getting 126 assets through copy, design, QA and three reviewers with one named owner per stage is production management, and it is where most nine-flow programs quietly become five-flow programs. Tools like LaunchSign exist for the second half: the journey is laid out as one canvas, every message and landing page in it carries its own owner and status, and reviews run as rounds on a single version so nobody approves an asset that has since changed.
FAQ
Do I need a specific ESP for this?
No. Every major ESP, including Klaviyo, Braze, Salesforce Marketing Cloud, HubSpot and Brevo, can execute all nine of these flows. The constraint on a nine-flow program is production capacity and review throughput, not platform capability. If your program has three flows, the reason is almost never that the tool could not do the other six.
How many messages should each flow have?
Between two and six, depending on the decision the person is making. Abandoned cart for an impulse product works at two. A considered purchase over $300 usually needs four or five, because the objections are sizing, returns and comparison rather than urgency. Four is a planning average for counting production volume, not a target for any individual flow.
Should these flows be email and SMS, or email only?
Start email only, and add SMS to the three flows where timing changes the outcome: abandoned cart, replenishment and VIP. Adding SMS to all nine roughly doubles your asset count and your review load for a marginal gain on flows where nobody is waiting on a time-sensitive decision.
What if my order volume is too low for nine flows?
Below roughly 500 orders a month, several of these flows will not have enough entries to evaluate. Replenishment, price drop and VIP are the first to fall below useful volume. Build the first five, measure for a full purchase cycle, and add the rest when the segment sizes justify the production cost.
How do I know whether to fix an existing flow or build a new one?
Compare the size of the audience entering each. A welcome series reaching 4,000 new subscribers a month with a weak first message is worth more attention than a price drop flow that would reach 200. Fix before you build whenever the existing flow has the larger audience, because the production cost of an edit is a fraction of the cost of a new flow.
