Browse abandonment is not cart abandonment, and sending the same email costs you both
One of them is a decision that got interrupted. The other is a question somebody was asking. The reminder that works for the first reads as surveillance when you send it for the second.
Cart abandonment recovery works because the person had already decided. They chose a size, they chose a colour, they got as far as a cart, and then the bus arrived or the card was upstairs. The email is a favour: here is the thing you were in the middle of.
Browse abandonment is a different event wearing similar clothes. Somebody looked at a product. That is not an interrupted decision, it is a question — often "is this the one?", sometimes "what does this even cost?", frequently "no". Sending them the cart email, with its urgency and its countdown, answers a question they were not asking and tells them you were watching.
Not every view is a signal
The first job is throwing most of it away. A single product view is noise: people arrive from search, realise it is the wrong thing, and leave. Treating that as intent produces a flow that emails everybody about everything.
- Three or more views of the same product inside a session, or across two sessions in 48 hours.
- Or two or more products from the same category with more than thirty seconds on each.
- And no cart — the moment there is a cart, this is the other flow's job, and only one of them may send.
That threshold cuts the audience hard, which is the point. A browse flow that fires for 3 % of sessions and converts is worth more than one that fires for 40 % and trains people to ignore you.
Wait longer than you want to
Cart recovery is fast: the first message inside an hour, because the decision is still warm. Browse is the opposite. Send at four to six hours, or the next morning if the browsing happened late.
One message. Possibly a second after three days if they looked again and still have not bought. There is no third: at that point the answer was no, and a third email is how you find out they have a spam button.
Write it as a question, not a recovery
The cart email can say "you left something behind", because they did. The browse email cannot, because they did not.
- Lead with the answer, not the nudge. The thing they were probably trying to work out: what it is made of, how it fits, what it costs delivered.
- Show two or three alternatives. Browsing without adding to cart often means "nearly right". The alternatives are the value.
- Say nothing about stock or timers unless it is true and verifiable. Manufactured urgency on a browse is the fastest way to be marked as spam.
- No discount. You have no evidence of price resistance. You have evidence of a question.
You still need permission, and the tracking has to be honest
A browse flow depends on knowing who was browsing, which means identifying a person before they have bought anything. That is fine and it is normal — and it is also exactly the processing that consent rules are about.
- Only send to people with marketing consent that has a recorded source and a timestamp. A browse flow to somebody who never subscribed is the clearest possible complaint.
- Respect the analytics choice. If somebody declined tracking, they do not enter this flow, however tempting the intent signal.
- Expect the identity to be partial. A person on a phone and a laptop is two trails until one of them identifies; a flow that assumes otherwise sends duplicates.
Measure it against nothing, honestly
Browse flows flatter themselves. The people who trigger one were already interested, so attributing their purchase to the email overstates it enormously. Hold back 10 % as a control and leave it there permanently. The number you want is the difference, and it is usually smaller and more durable than the dashboard's.
If the difference is not there after a quarter, the flow is not earning its unsubscribes. Turn it off. A browse flow is the most switchable-off automation you own, and shops keep them running for years on a headline number that never had a control behind it.
Sources and further reading (3)
- EDPB — Guidelines 05/2020 on consent under the GDPR
- EDPB — Guidelines 2/2023 on Article 5(3) of the ePrivacy Directive
- Google — Email sender guidelines (complaint thresholds)
Checked on 23 September 2026. Provider prices, mailbox rules and legal guidance change — verify anything you plan to act on.
Browse and cart, told apart properly
Auralata sees the page views and the cart from the same shop, so a browse flow steps aside the moment a cart exists — and never enters somebody who declined tracking.