Journey Optimizer custom outbound channels reach GA

Custom outbound channels in Adobe Journey Optimizer are now generally available, with an availability date of 2 October 2026. If your brand talks to customers on a messaging service Journey Optimizer doesn't cover natively, you can now build that channel yourself and use it with the same tooling as email or SMS. A handful of other changes are flagged as coming soon, and they matter mostly if you run Direct Mail, custom SMS providers or large journey estates.

Key takeaways

• Custom outbound channels moved from Limited Availability to General Availability, with an availability date of 2 October 2026.
• The GA release adds Decisioning in the channel payload, business rules on custom channels, and selection in API-triggered campaigns.
• Direct Mail, custom SMS authentication, journey and Loyalty items sit under Coming soon. Adobe says they are subject to change, so plan for them but don't promise dates.
• Fewer step events for wait and event activities can change what your journey reports count. Check before anyone notices a drop.
• Pilot one custom channel with one use case, test consent and preview first, and use a partner for provider integration and governance rather than for button-clicking.

What custom outbound channel General Availability actually changes

The feature isn't new. Custom outbound channels were available to a limited set of environments before, and the October 2026 notes move them to every environment. The headline is access. If you were waiting for the capability to leave Limited Availability before putting it in a plan, that wait is over.

The idea is simple. An administrator uses a no-code Channel Builder to describe an outbound channel that runs over HTTP. The notes name WeChat, Kakao Talk and Messenger as examples, and also mention a proprietary provider. After that, the channel shows up across campaigns, journeys and orchestrated campaigns. Adobe lists the same capabilities native channels get: personalization with the expression editor, content experimentation, preview and proof, out-of-the-box reporting, and consent and governance enforcement.

Three additions come with GA:

1. Journey Optimizer Decisioning can be used in the custom channel payload through the Personalization Editor, the same way it works in code-based experiences.
2. Business rules can be applied to custom channels, as they already can on native channels.
3. Custom channels can be selected in the channel list for API-triggered campaigns, which wasn't possible before.

The second and third items are the ones that remove workarounds. Teams that pushed messages to a niche provider through a custom action often did so because the channel couldn't take part in rules or API triggers. Some of those workarounds can now be retired, although you should read the detailed documentation before assuming parity with your exact setup.

Who needs to act on custom channels, and who can wait

The change matters most to three kinds of team.

The first is any team in a market where a regional messaging app carries more customer attention than email. If WeChat or Kakao Talk is where your customers actually respond, a native-feeling channel inside the same orchestration tool beats a side system with its own audience lists.

The second is teams already running a custom action to reach a provider. They have a working path, but it likely sits outside consent enforcement, outside the standard reporting, and outside the preview workflow. Moving to a custom channel could bring all three under one roof. Whether it's worth the migration depends on volume and risk. A low-volume transactional nudge may not justify the effort. A promotional channel that touches consent law probably does.

The third is platform owners who get asked every quarter whether Journey Optimizer can reach some new provider. GA gives them a repeatable answer.

Who can wait? Teams whose customer contact is covered by email, push and SMS, and who have no pressing provider on the horizon. You don't need a custom channel to justify the release. Read it, note it, and move on.

One caution. The notes say the channel is configured by administrators. That means the people who build journeys may not be the people who define the channel. Agree early on who owns the channel definition, who approves changes to it, and who gets paged when the provider's endpoint misbehaves.

How to test a custom channel before it carries real traffic

Treat the first custom channel like a new vendor integration, because it is one. A sound sequence looks like this:

1. Pick one provider and one use case. A single confirmation message is better than a full nurture programme.
2. Build the channel in a non-production sandbox and send to internal test profiles only.
3. Use preview and proof with realistic profiles, including ones with missing attributes, long names and non-Latin characters if your market needs them.
4. Confirm consent behaviour. Create a test profile that has opted out and check that the message does not go out. The notes say consent and governance enforcement applies, but verify it on your own policies rather than inferring it.
5. If you plan to use Decisioning in the payload, test the fallback. What does the customer see when no offer qualifies?
6. Check the out-of-the-box reporting against what the provider's own logs say was delivered. Gaps will show you where the provider's delivery receipts don't map neatly to the channel's reporting.
7. Only then move to a small live audience and watch it.

The API-triggered campaign path deserves its own test. If another system fires the campaign, confirm that the trigger payload contains what the channel needs, and that failures upstream don't silently produce an empty message.

Keep a record of what you tested. The next channel will take half the time if the checklist already exists.

Direct Mail and custom SMS: the Coming soon items

The notes also list items under Coming soon. Adobe states that entries there are expected in the coming days or weeks and that the information is subject to change. So read them as direction, not as commitments.

For SMS, the notes describe more flexibility in custom authentication headers when connecting a provider's OAuth setup. You can configure where the token sits on outgoing messages and how the token request itself is formatted. This is for teams whose SMS provider doesn't fit the standard pattern. If you currently have a provider you can't connect cleanly, ask your security team and the provider whether the new options fit, then test in a sandbox when the feature appears in your environment.

For Direct Mail, two changes are listed. Files can be split into multiple parts automatically when they exceed roughly 20 GB, or manually by choosing a target file size in the file routing configuration. And the audience limit rises from 3 million to 100 million profiles, which the notes tie to avoiding file-creation errors on large audiences. If you run print campaigns at scale, the practical question is your print vendor. A bigger audience limit does nothing for you if the vendor can't accept the files in the sizes you'll now produce. Ask them what they handle before you plan a larger send.

A few other items sit in the same section. Dynamic token substitution for Experience Manager fragments adds a tokenSubstitution attribute, which defaults to true and keeps the existing behaviour. Setting it to false makes personalization inside the fragment's fields resolve directly, without a token map in the reference. Adobe says this is Limited Availability for a set of organizations, so contact your Adobe representative for access. AEM Managed Services content fragments are also listed as supported in Decisioning when managing decision items. And Loyalty challenges gain per-member completion deadlines, using the option Within a number of days after opt-in under Completion requirements.

Journey changes: step events, deactivation and the AI skills

The journey section is a mix of housekeeping and new tooling. Start with the one that can bite.

Step events will no longer be generated for wait activities and event activities when the profile wasn't actually processed at that activity. If any dashboard, export or alert of yours counts step events, the numbers can change once this lands. That isn't a bug. It's fewer rows that never described a real pass. But a report that compares this month with last month could show a drop that looks like an audience problem. Find the queries that depend on those events now, and note the change date in the report so nobody chases a phantom.

A new Inbound Activity Deactivation activity on the journey canvas lets you remove a profile from up to five inbound activities or experiences. The point is decoupling: inbound disqualification no longer has to be tied to journey exit. A realistic case is a customer who completes a purchase mid-journey. You can pull them out of an on-site experience without ending the whole journey. Map your current inbound setups first, so you know where this would simplify things rather than add another thing to maintain.

On the AI side, the notes list several additions:

• An AI recommendation card on the home page when a journey alert fires, for Journey Custom Action failure and Journey Anomaly Detected alerts. Selecting it opens the journey with the right rail already populated with the analysis.
• Journey creation with AI from the Coworker right rail, replacing the previous AI Assistant experience.
• A Hygiene Analysis skill in CX Coworker that scans active and draft journeys for broken configurations, silent failures, and decaying or unused assets, such as stale drafts, orphaned data sources and persistent custom action errors.
• A Business Performance Analysis skill that analyzes journey performance and gives optimization recommendations in chat.

The hygiene skill is the one most teams will get value from. Most mature instances carry a pile of old drafts and custom actions that fail quietly. Treat its output as a prioritized list to review, not a cleanup order. A person who knows the programme should decide what gets retired.

A rollout checklist with owners

Use this as a starting point and adjust for your team.

Note that the Coming soon rows depend on rollout reaching your environment. Check the release notes and your own sandbox before scheduling work against them.

Where a partner helps and where your team can manage alone

Your team can handle a lot of this without outside help. Reading the notes, annotating reports, running the hygiene scan, and piloting a first custom channel with a simple provider are all within reach of a competent in-house admin and a journey builder.

A partner earns its fee in a narrower set of places. One is the provider integration itself, when the endpoint has quirks, rate limits, or a delivery receipt model that doesn't map cleanly. Another is governance design: who may create channels, how consent is mapped per channel, how you audit it. A third is migration of an existing custom-action setup into a proper channel, where the risk lies in moving live traffic without a gap. And a fourth is regional compliance for messaging apps, where local rules on consent and sender identity vary and a mistake is expensive.

If you do bring one in, brief them on the use case and the constraints, not on the feature name. Our RFP template for Adobe Journey Optimizer gives you a structure for that, and you can look at Adobe Journey Optimizer partners to see who works in this area. Ask each candidate to describe a channel integration they've actually delivered, including what went wrong in testing.

What not to do with this release

Don't migrate every custom action into a custom channel because you can. Start with the one that has the clearest compliance or reporting gain.

Don't plan campaign calendars around the Coming soon items. Adobe says they are subject to change, and your environment may receive them later than someone else's.

Don't let the AI skills act as an approval step. A recommendation card or a hygiene finding is a lead for a human to check. It is not a verdict.

Don't skip the consent test because the notes say governance is enforced. Your policies are specific to you, and the only way to know the setup honours them is to try to break it with an opted-out profile.

And don't ignore the step-event change just because it's listed as housekeeping. Quiet changes to what gets counted cause the loudest arguments about performance a month later.

If you're still weighing whether Journey Optimizer suits a channel mix like yours, compare partners or browse the blog for related guides before committing to a build.

Sources

• Adobe Experience League: Journey Optimizer release notes — Release notes page, last updated 7 October 2026, covering the October 2026 updates for channels, journeys, integrations and Loyalty.