Close content performance gaps with data-driven updates

AI client comparing the original and updated page content side by side {modal="regular"}

Select to zoom.

Content performance gaps cost campaigns results, and closing them usually means switching between your analytics tool and your CMS. This walkthrough shows how to do it in a single AI session: surface campaigns with conversion gaps in Customer Journey Analytics, diagnose the cause, inspect the underperforming content in AEM, and apply updates without leaving your conversation.

Scenario details
CX Enterprise applications
Customer Journey Analytics, Adobe Experience Manager as a Cloud Service
Agentic tools
CX Enterprise MCP, AEM Content MCP Server
Audience
Campaign managers, content strategists, marketing operations
Prerequisite
MCP-compatible AI client, CJA access, AEM as a Cloud Service access

Each step shows one representative prompt and an example AI response. A More you can accomplish section follows for additional exploration in the same session.

Before you begin

Claude.ai

Connect both MCP servers as custom connectors. Add each one separately.

  1. Go to Settings > Integrations in Claude.ai.
  2. Select Add custom connector, enter a server URL, and select Connect.
  3. Sign in with your Adobe ID, then repeat for the second server.
table 0-row-2 1-row-2 2-row-2
Server Endpoint
CX Enterprise MCP https://cx-enterprise.adobe.io/mcp
AEM Content MCP Server https://mcp.adobeaemcloud.com/adobe/mcp/content

Full setup: Claude.ai Custom Connectors documentation

ChatGPT

Connect both MCP servers using ChatGPT Developer Mode (Pro, Plus, Business, Enterprise, or Education plan required). Add each server separately.

  1. Enable Developer Mode in ChatGPT Settings.
  2. Go to Settings > Integrations and select Add custom connector > Remote MCP server.
  3. Enter a server URL, select Connect, and sign in with your Adobe ID.
  4. Repeat for the second server.
table 0-row-2 1-row-2 2-row-2
Server Endpoint
CX Enterprise MCP https://cx-enterprise.adobe.io/mcp
AEM Content MCP Server https://mcp.adobeaemcloud.com/adobe/mcp/content

Full setup: ChatGPT MCP documentation

Other AI clients

Using Gemini, Microsoft Copilot, Cursor, Claude Code, or another MCP-compatible environment? Connect to both MCP servers using these endpoints:

table 0-row-2 1-row-2 2-row-2
Server Endpoint
CX Enterprise MCP https://cx-enterprise.adobe.io/mcp
AEM Content MCP Server https://mcp.adobeaemcloud.com/adobe/mcp/content

Full setup instructions for all supported clients: Connect to your AI client

NOTE
Sign in with your Adobe ID when prompted and select the IMS organization linked to your CJA and AEM environments. Choosing the wrong organization is the most common source of authentication errors.
On first connection, your AI client may ask you to select an IMS organization or specify a sandbox. Once that context is set, the MCP server uses it for the rest of the session.
Some tools prompt for your approval before they execute. Review the request and approve or decline. No action is taken without your confirmation.

Step 1: Find campaigns with a conversion gap

Use CJA to surface campaigns where click-through is strong but conversion rate is low. This pattern (high intent, low completion) typically points to a content or experience problem on the landing page.

Which campaigns have strong click-through but low conversion in the last 30 days?
See an example response

AI client surfacing campaigns with high click-through but low conversion from CJA {modal="regular"}

Select to zoom.

Step 2: Diagnose the root cause

Follow up to understand what is driving the gap. Ask whether the drop-off is concentrated on a specific device type, audience segment, or content interaction.

What's causing the conversion drop-off, is it device, segment, or content?
See an example response

AI client diagnosing the conversion drop-off by device, segment, and content factors {modal="regular"}

Select to zoom.

Step 3: Review the content in AEM

With the underperforming campaign identified, pull the landing page from AEM in the same session. Seeing what the page currently says is the starting point for understanding what to change.

Show me the Bali Surf Camp page.
See an example response

AI client showing the current content of the landing page from AEM {modal="regular"}

Select to zoom.

Step 4: Get targeted recommendations

Ask your AI client to connect what the data showed with what is on the page. The AI reasons across both sources to identify which content sections are likely causing the drop-off and what to change.

Which content sections are underperforming, and what changes would you recommend?
See an example response

AI client identifying underperforming content sections and recommending specific changes {modal="regular"}

Select to zoom.

Step 5: Apply and review the changes

Ask your AI client to create an optimized version of the page based on the recommendations and summarize what changed and why.

Create an optimized version of the Bali Surf Camp page and summarize the proposed changes.
See an example response

AI client creating an optimized version of the page and summarizing the changes {modal="regular"}

Select to zoom.

CAUTION
Review the full summary of proposed changes before confirming. The AEM Content MCP Server will write changes to your AEM environment. Pages remain in their published state until you explicitly republish.

What you accomplished

You connected Customer Journey Analytics and AEM in a single AI session and moved from campaign data to deployed content changes without switching tools. You identified campaigns with conversion gaps, diagnosed the root cause, inspected the landing page, received targeted recommendations grounded in both data and content, and applied the changes in the same conversation. This shortens the feedback loop between analytics insight and published content, and scales to any number of underperforming pages in the same session.

More you can accomplish

With CJA and AEM connected in the same session, you can cover the full cycle from identifying problems to shipping fixes. Expand a scenario below to see prompts you can try.

Find the content that's holding back performance

High traffic with low engagement signals a content problem, not a traffic problem. These prompts help you surface specific pages and patterns that need attention before a campaign deadline forces the issue.

Prompts

code language-none
Which campaigns have the highest traffic but lowest conversion rate this quarter?
code language-none
Which pages have a high bounce rate but also high traffic?
code language-none
Compare engagement rates for landing pages across email and paid social campaigns.
code language-none
Find AEM pages linked from active campaigns that haven't been updated in over 60 days.
Fix what the data is telling you to fix

Once you know what’s underperforming, make targeted changes grounded in what the performance data revealed. These prompts let you update specific sections based on the diagnosis.

Prompts

code language-none
Update the CTA on the [page name] page to better match the campaign audience.
code language-none
Rewrite the hero headline on the [page name] page to address the mobile drop-off.
code language-none
Add a trust signal to the [page name] page above the conversion form.
code language-none
Which pages updated in this session still need to be published?
Ship improvements before the next campaign

Changes made mid-session can pile up quickly. These prompts help you review what is ready, group updates for review, and promote cleanly before a campaign goes live.

Prompts

code language-none
Show me all pages updated in this session that are still unpublished.
code language-none
Create a launch with all changes from this session for review before publishing.
code language-none
Give me a summary of all changes made in this session.
code language-none
Publish all confirmed changes and share the updated URLs.

Further information

Resource
What you will find
CJA MCP Server in AI Registry
CJA MCP Server tools and availability
AEM Content MCP Server in AI Registry
AEM Content MCP Server tools and availability
recommendation-more-help
cx-enterprise-agentic-tools-help