When you use Adform MCP Server, your AI application works with data and capabilities available through Adform FLOW. Adform MCP Server provides the requested Adform data, while the AI application processes that data to respond to your prompt. As a result, the response you receive also depends on your prompt, the skills you use, and how your AI application interprets and presents the information returned through MCP.
This article helps you identify whether an issue is related to Adform MCP Server or your AI application and where to get help.
Tip
If you can't establish or authenticate the MCP connection, see Troubleshoot MCP Connection.
Important
Your use of Adform MCP Server is subject to your existing agreement with Adform and the MCP Access Notice. MCP provides read-only access and does not extend your existing Adform Platform access rights. You are responsible for ensuring that you are authorized to access and use Adform data through MCP and your chosen AI application, and for reviewing and validating AI-generated outputs before relying on them.
For terms, responsibilities, and restrictions, read the MCP Access Notice.
When you prompt your AI application, it decides what information it needs and which available MCP tools to use. Adform MCP Server retrieves the requested Adform data and returns it to the AI application. The AI application then interprets the data and prepares the response you see.
Adform MCP Server can also return forecasts, recommendations, and other insights that originate in Adform FLOW. However, additional analysis, reasoning, conclusions, or suggestions generated by the AI application are produced outside Adform.
Because the final response involves both Adform MCP and your AI application, first identify which part of the process might be causing the issue.
First, check whether the information available through Adform MCP matches what you can access in Adform FLOW. This can help you determine whether the issue relates to your Adform access or to how your AI application processes the data.
Adform MCP Server access depends on your Adform FLOW permissions and OAuth scopes. The MCP server returns only data your user account is authorized to access in Adform FLOW.
If you can't access an expected business account, advertiser, campaign, or other Adform data:
-
Make sure you're working with the expected Adform business account.
Note
With Adform MCP Server, you can work with one business account per MCP connection.
-
Check whether you can access the expected information in Adform FLOW.
-
If your access in Adform MCP differs from your access in Adform FLOW, contact Adform Support (technical@adform.com).
If the information returned by your AI application differs from what you see in Adform FLOW, first determine whether the difference is in the underlying Adform data or in how your AI application interpreted or presented that data. Some discrepancies can occur because of data aggregation or reporting timing.
If the underlying Adform data is correct, but your AI application interprets, summarizes, visualizes, or presents it differently than expected, check your AI provider's documentation and support resources.
If the underlying Adform data itself appears incorrect or incomplete, contact Adform Support (technical@adform.com) and include:
-
The business account, advertiser, campaign, or other relevant entity
-
The data or metric you expected
-
What you see in Adform FLOW
-
What your AI application returned
-
The time period used in the request
-
Your original prompt
-
Screenshots, where helpful
This gives us the information needed to determine whether the issue originates from the Adform data or from how the AI application interpreted it.
Not all Adform data and functionality is available through MCP. If your AI application can't retrieve specific information, the requested data might not be exposed through the available MCP capabilities, or the AI application might not have selected an appropriate MCP tool for the task.
To troubleshoot, try the following:
-
Make sure that Adform MCP Server supports the action you're trying to accomplish. Consider Adform MCP server capabilities and limitations.
-
Make your request more specific. Include relevant details such as the advertiser, campaign, metric, and time period.
-
Ask your AI application what Adform data or MCP capabilities it used to produce the response.
-
Check whether an Adform MCP skill is available for the workflow.
-
If the information still isn't available, contact Adform Support (technical@adform.com) with an example of what you're trying to retrieve and the expected result.
Note
Adform skills can help AI applications work more consistently with Adform MCP, but they don't extend your FLOW permissions or change the connected business account.
If Adform MCP Server can retrieve the expected data, but the response is incomplete, unclear, or different from what you expected, the issue might relate to your prompt or how your AI application processes and presents the returned data.
If you use Adform MCP skills, the instructions in those skills can also influence how your AI application completes the task. While skills provide reusable instructions for common workflows, they don't replace a clear prompt.
Note
Skills help guide AI applications by providing task-specific context and instructions. However, skills don't guarantee a particular outcome, as it still depends on the prompt, the data retrieved through MCP, and how the AI application interprets and applies the skill instructions. Make sure to review and verify the AI-generated responses before use.
To improve the response:
-
Clearly state what you want to know or accomplish.
-
Include relevant context, such as the advertiser, campaign, currency, dates, KPIs, or reporting metrics.
-
Specify the desired AI response output format when relevant.
-
Consider breaking complex requests into smaller tasks.
-
Use an Adform MCP skill when one is available for the workflow.
-
Ask your AI application to include the supporting Adform data it used to produce the response.
-
Check what AI model your AI application is using for the task, and consider if it's optimal for your desired outcome.
Note
Adform does not provide or manage third-party AI applications. You are responsible for any costs charged by your AI application provider, including subscription fees and token consumption.
Issues can originate from Adform MCP Server, your AI application, or how the two work together. You might notice the issue yourself, or your AI application might indicate that it can't retrieve or process the expected Adform data.
Use the guidance below to determine where to get help. If you're unsure where the issue originates, contact Adform Support and we'll help you identify the appropriate next step.
Contact Adform Support (technical@adform.com) if the issue concerns:
-
Authentication or authorization with Adform MCP Server
-
Access to Adform MCP or Adform FLOW data
-
Missing or unexpected business account access
-
Adform permissions
-
Adform data returned through MCP that appears incorrect or incomplete, compared to data in Adform FLOW
-
Adform FLOW data or functionality that you expect to be available through MCP but can't access
-
Adform MCP Server availability or operation
-
Errors indicating that Adform MCP can't retrieve or return the expected data, for example, because of an Adform backend or API issue
When reporting an issue, provide the AI application you're using, your prompt, the expected result, the result you received, relevant screenshots, and details about the affected Adform entities. This helps separate an Adform-data issue from an AI-application issue.
Refer to your AI provider's documentation or support resources if the issue concerns:
-
Your AI application settings, configuration options, or interface behavior
-
How to write or improve prompts
-
How the AI application selects or uses MCP tools
-
How the AI application interprets Adform data
-
How results are formatted, visualized, or presented, such as charts or summaries
-
Unexpected AI-generated analysis, reasoning, conclusions, or suggestions
-
Provider-specific MCP functionality
-
Features, limitations, subscriptions, or usage limits of the AI application, assistant, or agent
For more information, see your provider's official MCP documentation:
-
ChatGPT: Model Context Protocol documentation and Developer mode and MCP apps. They cover how ChatGPT and Codex use MCP and provide separate guidance for developer mode and custom MCP apps.
-
Claude: Claude Model Context Protocol documentation. This explains Claude connectors, remote and local MCP servers, authentication, tools, resources, and prompts.
-
Microsoft Copilot: MCP in Copilot Studio. This covers MCP tools and resources in Copilot Studio and links to its MCP integration troubleshooting guidance.
-
Mistral: MCP Connectors documentation. This documentation covers custom MCP connectors, authentication, and connector permissions.
-
Other AI applications: Refer to the respective AI provider's official documentation.
Note
AI application providers might update their interfaces and MCP connection workflows. If the settings or options differ from what you see, refer to your AI provider’s official documentation for the latest MCP connection instructions.
If you aren't sure where the issue originates, contact Adform Support (technical@adform.com). We'll help identify whether the issue relates to Adform MCP, your Adform access or data, the connection configuration, or your AI application. If the issue relates to Adform, we'll continue investigating it. If it relates to your AI application, we'll direct you to the appropriate provider resources.