CJA MCP Read-Only Access and Auto-Generated Descriptions

Adobe's latest Customer Journey Analytics release notes are less about new data and more about who can touch what, and how easily people can read the result. The two items that deserve a decision this month are the new MCP Read-only Access permission, generally available since 6 October 2026, and generated descriptions for components and projects, generally available on 18 November 2026. Around them sit smaller Workspace changes, plus Conversation Insights and Adobe Brand Visibility, which need more thought before you commit to them.

Key takeaways

• MCP Read-only Access became generally available on 6 October 2026, and the old MCP Access item is now called MCP Full Access, so check who holds which.
• Generated descriptions for components and projects reach general availability on 18 November 2026. Treat the output as a draft to review, not finished documentation.
• Compare percentages (19 October 2026) and Concise labels (29 October 2026) are small changes that make reports easier to read, and both are quick to test.
• Conversation Insights began rolling out on 29 October 2026 after slipping from 22 September 2026, and its general availability date is still to be announced. Pilot it; don't promise it.
• Most of this is admin and analyst work a team can do alone. A partner earns their fee on data design for conversation data and on governance across many teams.

What the MCP Read-only Access permission changes

The Customer Journey Analytics MCP server lets approved users work with the product through tools. Until now the access model was blunt. You had MCP Access, and it opened everything. Adobe has now split that in two. MCP Read-only Access gives a user all the read-only tools but stops them creating projects, segments or calculated metrics. The original permission item is renamed MCP Full Access, and people who hold it keep every tool, including the ones that create, change or delete components. Adobe lists the read-only option as generally available from 6 October 2026.

The rename matters as much as the new item. If you granted MCP Access last quarter, those users now effectively hold MCP Full Access. Nothing in the notes says their rights shrank, which means they can still make changes through the server. That may be exactly what you intended for two senior analysts. It is probably not what you intended for a marketing manager who only wanted to ask questions.

The practical lesson is the usual one about least privilege. Someone who needs to look things up, summarise a report or pull a metric does not need the ability to delete a segment that three dashboards depend on. Read-only access removes that risk without removing the convenience.

For the exact list of tools in each tier, go to the permissions section of the Customer Journey Analytics MCP server documentation. Don't guess from the names.

Who needs to act on permissions first

Not every team has the same exposure. A small analytics group where everyone is a power user has little to fix. A large organisation where product managers, regional marketers and agency staff all hold product logins has more to review. The table below sorts the situations by urgency.

The last row is the cheapest to get right. Setting a read-only default before people ask for access is far easier than taking rights away later.

Generated descriptions for a crowded component library

Anyone who has inherited a Customer Journey Analytics data view knows the problem. There are 400 metrics, 90 segments and a pile of calculated metrics with names like Final v2 NEW. Nobody is sure which is current. Adobe's answer is automatic description generation for dimensions, metrics, calculated metrics, segments and date ranges. You can generate one description or many at once. Projects get the same treatment when you save them or from Project info and settings. Rollout started on 29 October 2026, and general availability is listed as 18 November 2026.

This is useful, but treat it as a first draft. A generated description can say what a component appears to contain from its definition. It cannot know that your Qualified Lead metric excludes a particular campaign for a contractual reason, or that a segment was retired in spring. That knowledge lives in people's heads, and it is the part worth writing down.

A sensible way to use it:

1. Pick one data view and filter to the components people use most, not the whole library.
2. Generate descriptions in a batch.
3. Have the component owner read each one and fix anything wrong or vague.
4. Add the business rule the generator could not know, such as a definition, an exclusion or an owner.
5. Only then widen to the next data view.

The temptation is to generate everything and call the documentation job done. Resist it. A confident but wrong description is worse than a blank field, because people trust it.

Compare percentages and Concise labels in Workspace

Two smaller changes will help report consumers more than their size suggests.

Compare percentages applies to the Journey canvas. When you compare a journey with a prior period, you can switch the option on and see the change in percentage points for each node, arrow and fallout. Adobe's point is that this keeps a retention or fallout comparison accurate even when the number of people in the journey differs between the two date ranges. Raw counts can mislead here. If a campaign doubled traffic, a node with more people may actually be converting worse. Percentage-point change shows that. It became available on 19 October 2026.

Concise labels deals with a more everyday irritation. Legends in a chart with several metrics, segments and date ranges tend to repeat the same names in every label, and the part that differs is only visible on hover. The new setting strips out whatever every label shares, whether metric, segment, dimension item or date range. It is off by default. It supports area, bar, combo, donut, histogram, horizontal bar, line and Venn visualizations, including stacked versions, and became generally available on 29 October 2026.

Neither needs a project. Open two or three of your most-viewed Workspace dashboards, switch the options on, and see whether a senior reader gets to the point faster. If so, update the templates your team copies from.

Conversation Insights and Adobe Brand Visibility: plan, don't rush

Two items in the notes touch AI, and they carry different levels of certainty.

Conversation Insights brings unstructured chat data into Analysis Workspace. The idea is to report on LLM-powered browsing and buying experiences. You collect prompts, responses and agent metadata from conversational agents, either your organisation's custom agents or Adobe Brand Concierge, using Web SDK. Then you analyse intent, tone and sentiment, and tie conversations to outcomes such as conversion and engagement. Rollout started on 29 October 2026, moved from a planned 22 September 2026, and general availability is shown as to be determined.

Take that slip as a signal. A date that has already moved once, with no general availability date yet, suggests you should pilot rather than build a plan that depends on it. Pick one conversational agent, one property and one question, perhaps whether people who ask about pricing go on to convert. Check the data you collect against your privacy and consent obligations before anything goes into production. Chat transcripts can contain personal details that no one meant to share.

Adobe Brand Visibility integration is a different kind of item. It lets you connect Adobe Brand Visibility with your Customer Journey Analytics data, so you can measure how AI-driven discovery turns into website engagement and business outcomes. Adobe lists it as generally available from 8 October 2026. If your team already uses Brand Visibility, the integration is worth a look. If not, there is nothing here that forces you to start.

Coworker in Analysis Workspace and Content Analytics

Coworker shows up twice. The first item, CX Enterprise Coworker, lets people use Coworker Chat to ask questions about content in Content Analytics. Adobe's examples run from specific questions, such as which images drove the most revenue last quarter, to broad ones about what content to prioritise. Rollout started on 14 October 2026, with general availability on 3 November 2026.

The second is an in-product experience for Coworker, a chat panel inside Analysis Workspace. It uses the project you have open as context, so you can ask about your analysis without explaining what it contains. Adobe says you can summarise the project, find insights and run root cause analysis, and also build and edit the project with natural-language prompts, such as adding panels and tables, breaking down data and changing visualization types. Check the Adobe documentation for the full list of actions and its availability dates.

A natural-language layer is only as good as the data model under it. If your dimensions are named inconsistently, or two metrics mean nearly the same thing, the answers will carry that confusion. That is a good reason to do the description work above first. Clean, described components help human analysts and AI assistants alike.

It also changes who asks the questions. A marketer who would never have opened Workspace may now try the chat panel. Decide in advance who is allowed to, and what you want them to do with the answer. A summary is fine for a meeting. A figure pasted into a board deck should be checked by an analyst first.

Rolling it out and measuring whether it worked

None of this needs a big programme. A sequence of small steps, each with an owner, will do.

Measuring is simple enough that it should not be skipped. For permissions, count how many people hold full access before and after. For descriptions, track the share of your most-used components with a reviewed description, and watch whether support questions of the form which metric should I use fall. For the Workspace readability changes, ask three regular report readers whether they find answers faster. For the pilot, define in advance what a useful result looks like.

Where a partner helps, and where you can manage alone

Most of the items above are admin settings or Workspace options. Auditing permissions, running description generation and enabling Compare percentages or Concise labels are afternoon jobs for someone who already administers the product. Paying a consultant to do them is hard to justify.

A partner becomes useful when the work reaches past one team. Designing the schema and datasets for conversation data, so that prompts, responses and agent metadata line up with the rest of the customer journey, is real design work. So is setting a permission and naming model across business units, or deciding how consent applies to chat transcripts. Those decisions are easier to get right once than to fix after six months of data.

If you do bring someone in, write the brief around outcomes: what you want to measure, which data you will connect and who will own the result. The Customer Journey Analytics RFP template gives you a starting structure. To see who works in this area, browse the Customer Journey Analytics partners page, or use find partners if you want to filter by need.

A few things to avoid:

• Don't leave everyone on MCP Full Access because it was the default.
• Don't publish generated descriptions without a human read.
• Don't build a roadmap around a feature that has no general availability date.
• Don't switch on every new setting at once, or you won't know which one confused your readers.

The common thread is control. Decide who can change things, write down what your data means, and make the reports easier to read. Do those three well and the rest of the release is a bonus.

Sources

• Adobe Experience League: Current Customer Journey Analytics release notes — Release notes for October to December 2026, last updated 6 November 2026.