Journey exit criteria and goals

What it does

By default a customer leaves a journey when they finish the last tile. Exit criteria let you pull them out early — when a trigger fires or when they join a segment — and a journey goal defines what counts as success and measures how many people got there.

Key facts

Three configured ways out beyond finishing: exit on a trigger, exit by segment membership, and an end date after which nobody new enters.
Exit on a trigger removes the customer immediately, from wherever they are in the flow, including mid-wait.
Exit by segment is a suppression segment and works anywhere in the journey. This is different from an exclusion segment at the start tile, which only prevents entry and does nothing for people already in flight.
An end date stops new entries but lets people already in the journey finish their remaining steps.
The Stop button is not an exit criterion — it halts the journey outright. No new entries, and everyone in flight stops where they are with remaining steps skipped.
A stopped journey cannot be restarted. If the journey has multiple versions, each version has to be stopped individually.
Journey goals are defined by a trigger plus a target percentage of the audience, for example "50% complete a purchase".
The goal is what determines the winner of an A/B test and what feeds channel optimisation, so it is not purely a reporting field.
Goal completion is measured while the customer is in the journey, so a goal reached after they exit doesn't count.
Analytics keep moving for days after a journey stops, because messages already delivered continue to be opened and clicked.

When to use / skip

Set an exit criterion on any journey where the message stops being appropriate once the customer acts. Abandoned cart is the obvious one — exit on purchase completed, or you carry on nagging someone who has already paid. Renewal chasers, unpaid invoice reminders, and event registration pushes are all the same shape.

Skip exit criteria on genuinely informational journeys — a welcome series, a product announcement — where finishing the sequence is the point and there is no action that makes the remaining messages wrong.

Goals are worth setting on anything where someone will later ask "did it work". They cost nothing and they are the only place the journey states its own success criterion. On journeys with an A/B test or channel optimisation they are not optional in practice: without a goal there is nothing for the optimisation to optimise against.

Configuration decisions

Which trigger, if any, means "stop talking to this person" — and whether that trigger already exists or needs building.
Whether you need a suppression segment as well as an exclusion segment, and whether it is one global segment or a per-journey one.
Whether the journey has a hard end date, particularly for anything tied to an event, an offer window or a regulatory deadline.
The goal trigger and the target percentage, agreed with whoever will report on the campaign rather than guessed during build.
What happens to people caught mid-journey on the end date: finish naturally, or exit them with a suppression segment.
Who has the authority to press Stop in production, given it cannot be undone.

Gotchas

Exclusion at the start and exit by segment are different mechanisms and clients conflate them. Adding someone to the start exclusion segment does nothing for the 4,000 people already mid-journey.
Stop is irreversible. There is no pause-and-resume equivalent for a stopped journey, and on a multi-version journey you have to stop each version separately — miss one and messages keep going.
A goal reached after the customer exits is not counted, so an aggressive exit criterion can make a successful journey look like it failed.
Analytics continue to shift for several days after stopping. Do not report final numbers the same afternoon.
Exit on trigger fires immediately, including on people mid-wait who were about to get a genuinely useful message. Check that the exit and the content actually conflict before wiring it up.
An end date without an exit plan leaves people finishing a sequence into a period when the offer no longer exists.

Consultant notes

Ask "what makes this message wrong to send?" for every journey in the backlog. The answer is the exit criterion, and about half the time nobody had considered it.
Demo the exit-on-purchase case. It is the clearest example of the platform doing something a batch email tool cannot, and it usually justifies the project on its own.
Set goals on every journey during build even if reporting is out of scope. Retrofitting them means a new version and a broken comparison.
Before go-live, agree in writing who can stop a live journey and what the escalation path is. The Stop button gets pressed in a panic and cannot be taken back.
Check the suppression segment refresh cadence — a suppression segment that refreshes nightly is not an emergency stop, and clients assume it is.

Worth another look if journey pause/resume behaviour changes, or if goal measurement starts counting post-exit conversions.