Browsing, saving an item and starting checkout give a retailer different signals about what a visitor might want. Those signals can guide an automation workflow, but they do not tell us exactly what someone intends to do.
Here is a basic example using three visitors to the same online shop. It is an illustration of the workflow, not a client case study or a claim about expected sales.
Three visitors, three possible messages
- Visitor A browses four product pages without saving an item or starting checkout. If the visitor is eligible for remarketing, a product explanation or a useful category guide could help them decide whether the shop has what they need.
- Visitor B adds an item to a wishlist but does not purchase. A reminder about the saved item, its availability or a relevant alternative could be useful.
- Visitor C adds five items to a basket and starts checkout without completing it. A reminder that makes returning to the basket easy may be more helpful than another general brand ad.
I would treat those as working hypotheses. A wishlist can be a shopping list, a gift idea or something someone intends to revisit months later. A checkout can stop because of delivery charges, a payment problem or a distraction. An ad cannot resolve every one of those problems.
Define the rules before switching it on
Give each segment an explicit entry condition and an exit condition. For example, a completed purchase should remove that basket from the abandoned-checkout workflow. If someone qualifies for several segments, decide which message takes priority so that the workflow does not bombard them with conflicting prompts.
Keep the product information and stock status current. Set a sensible contact frequency, an expiry period for each audience and a route for someone to opt out. Only activate channels and audiences where the relevant consent, permissions and platform requirements are met. A website visit alone does not authorise every form of follow-up.
I would not make a discount the automatic response to every abandoned basket. It costs margin, and the customer may have purchased without one. First consider whether the message can answer a practical question about delivery, returns or payment.
Check whether it helps
Record which rule triggered each message and check that purchases, exclusions and opt-outs reach the workflow reliably. Start with a small test and inspect real examples of the messages it produces.
To evaluate the commercial result, compare eligible customers assigned to receive the workflow with a suitable holdout group where feasible. Look at incremental purchases and contribution after advertising and discount costs, alongside opt-outs and complaints. Platform-reported conversions alone do not establish that the workflow caused the sale.
This is one use of marketing automation. The same discipline—clear triggers, useful actions and reliable exits—also applies to email and other permitted channels. The data checks behind the workflow matter just as much as the message.