AI works best when it meets your team where they already are. For analytics teams using Adobe Analytics or Customer Journey Analytics (CJA) that's exactly what Adobe's Analytics MCPs make possible. It connects your analytics environment directly to the AI tools your team is already working in, so you can ask questions, pull data, and build reports without ever leaving your workflow.
MCP stands for Model Context Protocol. It's an open standard that lets AI tools like Claude or Cursor connect directly to external data sources and services rather than relying on you to copy and paste information into the chat. Instead of exporting a CSV and describing your data to an AI, MCP gives the AI a live connection to the actual system so it can query, explore, and build reports directly. In other words, it's AI meeting you and your team where you're already working, rather than asking you to change how you work to accommodate it.
Adobe offers two analytics MCP servers depending on which product you're working in: one for Adobe Analytics and one for CJA. Both work the same way conceptually: once connected, you can ask questions in plain English and the AI pulls live data back from your environment without ever leaving your coding tool.
What that looks like in practice is being able to say things like "show me daily sessions for the last 30 days broken down by marketing channel" and getting a real result without manually building a report in Analysis Workspace. You can also discover what dimensions, metrics, and segments are available in your environment, describe components you're unfamiliar with, apply filters, create or update segments and calculated metrics, and even have the AI build a Workspace project for you directly from the chat. It's worth noting that the MCP inherits whatever component settings are configured in your environment, including attribution models and persistence settings, so the results it returns reflect those configurations just as they would in Workspace.
An important thing to understand about how this works: the MCP mirrors your Adobe Analytics or CJA environment directly. It mirrors the same reporting interface you'd use in Workspace, the same components, the same data, the same structure. That means if the MCP builds something that looks off, it's typically not the tool making something up. It just needs more context or more specific instructions, which is exactly why what you do before you start querying matters so much.
Set your data view or report suite at session start. Both the Adobe Analytics and CJA servers let you set up session defaults so you're not specifying them on every request. For CJA this is your data view ID; for Adobe Analytics it's your report suite ID and global company ID. Consider setting up your own skill or reusable prompt that handles session setup automatically, so it becomes a reliable part of your workflow.
Know which components are approved or governed in your org. Adobe's MCP servers can search for the approved tag on metrics, segments, and calculated metrics if your company has them in place. This helps your team use better vetted components when creating reports in an AI tool instead of pulling something experimental or deprecated.
Confirm before you run. Instinctually we want to ask questions in whatever plain language we commonly use, but many metrics and dimensions in Adobe can have similar names with meaningfully different definitions. The difference between a metric that includes all sessions versus one that excludes certain traffic types, for example, can be significant. A workflow that holds up well is:
Ask. Confirm. Run. Validate.
Find the component, confirm it matches what you actually want before running, and after you get results back validate that the components and date ranges used were what you intended.
Understand the data lag. CJA has a normal processing lag, so same-day data should always be treated as preliminary. If results look off on a high-traffic day, data may still be catching up. A sudden flatline or unexpected drop in a time series is often a lag signal rather than a real business trend.
The Adobe Analytics MCPs don't replace Analysis Workspace; they get your analysis started faster without having to context-switch into a separate tool. At its core, it's AI meeting your team where you're already working, not the other way around. Getting the setup right and understanding the guardrails is what makes that possible.
Not sure on your next step? We'd love to hear about your business challenges. No pitch. No strings attached.
