Your attribution data, just one conversation away.
Every performance marketer knows the routine. A question comes up during the morning stand-up: Why did installs drop yesterday? Which partner is driving the most rejected traffic? How did last week’s campaign convert to first purchase?
Answering it often means opening a dashboard, selecting an app, adjusting the date range, applying filters, grouping results by partner, exporting a report, and repeating the process for another app.
The data is already there. Finding the answer shouldn’t take that much work.
Apptrove MCP simplifies the process by connecting your attribution data to the AI assistants you already use. With support for Claude, ChatGPT, Cursor, and Claude Code, you can ask questions in plain English and receive answers based on the reports available through your Apptrove account.
Whether you need a quick performance summary, a comparison across acquisition partners, or help investigating rejected installs, you can start with a question instead of building another report.
What Is Apptrove MCP?
MCP, or the Model Context Protocol, is an open standard that allows AI assistants to connect with external data sources and tools.
Apptrove MCP is Apptrove’s implementation of this standard. It runs on a hosted server at https://mcp.apptrove.com/mcp and gives compatible AI assistants read-only access to the Apptrove reports a user is already permitted to view.
Once connected, you can ask your AI assistant to retrieve attribution data, compare performance across reports, summarise results, and highlight changes worth investigating.
For example, you could ask:
- “Summarise yesterday’s installs and in-app events by partner, and flag anything that moved more than 20% from the previous day.”
- “Which three partners drove the most rejected installs last week, and what were the reasons?”
- “Compare Meta and Google Ads installs for October across all my apps.”
- “Build a funnel from registration to first purchase by media source for the last 30 days.”
Apptrove MCP retrieves the relevant information from the same reporting that powers your Apptrove dashboard. The assistant then uses those results to formulate an answer.
Your attribution data remains in Apptrove. The assistant requests the information it needs to answer your question, subject to your existing access permissions.
The goal is simple: spend less time navigating reports and more time understanding what your campaign data is telling you.
Turn Attribution Data Into Actionable Insights
The value of Apptrove MCP goes beyond retrieving individual metrics. It gives marketing, analytics, and technical teams a more direct way to investigate performance and connect information across reports.
Ask Questions in Plain English
You don’t need to remember which dashboard contains a particular metric or manually configure every report.
Ask a question in your own words, and your AI assistant can determine which Apptrove reports are relevant to the request.
For example, instead of building a report to understand daily performance, ask:
“Summarise yesterday’s installs, events, and revenue by partner. Highlight the biggest changes.”
You can start with a broad question and refine it as you explore the results.
Compare Performance Across Apps and Partners
Managing multiple applications can mean repeating the same reporting process for each one.
Apptrove MCP lets you request comparisons across the applications available through your account. You can also examine performance by partner, campaign, channel, or agency traffic.
For example, you could compare Meta and Google Ads installs across your apps or identify which acquisition partners are generating the most rejected installs.
This gives teams a faster way to investigate differences in performance without manually recreating the same report for every app.
Understand What Happens After an Install
Installs tell you how many users arrived, but they don’t explain what those users did next.
Apptrove MCP can retrieve in-app event performance, event-level funnels, customer reports, cohort tables, and retention data.
You can ask questions such as:
“Show the Day 1, Day 7, and Day 30 retention for users acquired from each partner in September.”
Or:
“Build a funnel from registration to first purchase by media source for the last 30 days.”
These questions help you explore the relationship between acquisition sources and post-install activity, using the reporting data available in Apptrove.
Investigate Fraud and Rejected Traffic
Acquisition volume alone doesn’t tell you whether a partner is delivering quality traffic.
With access to rejected-install and rejected-event data, Apptrove MCP can help you investigate partner quality and understand why traffic was rejected.
For instance, you could ask which partners generated the most rejected installs during the previous week and retrieve the associated rejection reasons.
You can also examine the validation rules configured for an application, helping your team investigate traffic-quality concerns with more context.
Troubleshoot Integration Issues Faster
When a partner reports missing conversions or an event doesn’t appear as expected, technical teams often need to inspect logs before they can identify the problem.
Apptrove MCP can retrieve samples of install, click, impression, event, postback, and uninstall logs for troubleshooting.
You could ask:
“Pull a sample of postback logs for this partner and tell me whether the postbacks are firing correctly.”
The assistant can retrieve the relevant log samples and summarise what was sent, received, or rejected, giving your team a more direct starting point for investigating integration issues.
What Can You Ask Apptrove MCP?
Apptrove MCP supports several areas of reporting and configuration. The exact information available depends on your permissions and the applications accessible through your connection.
| Reporting area | What you can retrieve | Example question |
| Performance summaries | Daily, multi-app, and dashboard-level installs, events, clicks, revenue, and uninstalls | “Summarise yesterday’s performance across my apps.” |
| Partners and campaigns | Partner, campaign, and channel breakdowns, including agency traffic | “Which partners generated the most installs last week?” |
| In-app events | Event-level funnels and event performance by source | “Compare registration-to-purchase conversion by media source.” |
| Users and cohorts | Customer reports, cohort tables, and retention | “Show Day 1, Day 7, and Day 30 retention by partner.” |
| Deep links | UniLink template and link performance | “How many installs and events did each UniLink template generate this month?” |
| Fraud and validation | Rejected installs, rejected events, and app validation rules | “Which partners had the most rejected installs, and why?” |
| Logs | Samples of install, click, impression, event, postback, and uninstall logs | “Retrieve postback logs for this partner for troubleshooting.” |
| Configuration | App settings, integrated partners and agencies, and SDK integration details | “Which partners are integrated with this app?” |
These capabilities allow you to move from a quick performance check to a more detailed investigation without manually assembling every report.
How Apptrove MCP Works
Apptrove MCP connects your AI assistant to Apptrove’s reporting capabilities through the Model Context Protocol.
The workflow is straightforward:
- Ask a question. Enter a request about your attribution data in your connected AI assistant.
- Identify the relevant reports. The assistant determines which Apptrove reports it needs to answer your question.
- Review the request. The assistant presents the request for your approval before it runs.
- Retrieve the data. Once approved, the request is sent to Apptrove MCP, which retrieves the permitted information.
- Get your answer. The assistant uses the returned data to summarise results, compare metrics, and explain relevant changes.
A single question may require information from multiple reports. For example, understanding partner quality could involve comparing a partner breakdown with rejected-install data.
Rather than manually opening both reports and comparing their results, you can ask the question directly. Your assistant coordinates the required requests, subject to the approval process.
The figures returned are drawn from the same reporting that powers your Apptrove dashboard, helping keep your analysis grounded in your account’s reporting data.
Built for the AI Tools You Already Use
Apptrove MCP works with MCP-compatible AI clients, including Claude, ChatGPT, Cursor, and Claude Code.
This means you can bring attribution reporting into the AI environment that already fits your workflow instead of adopting an entirely new interface.
Claude
Add Apptrove MCP to Claude as a custom connector using the hosted MCP server address. Authenticate with your Apptrove credentials when prompted, then start asking questions about your reporting data.
ChatGPT
Connect Apptrove MCP to ChatGPT through a compatible MCP connection. Once configured and authenticated, you can ask questions about your Apptrove data directly within your AI workflow.
Cursor and Claude Code
Developers can connect Apptrove MCP to Cursor and Claude Code using the MCP server address and the authentication method supported by their client configuration.
If you configure the connection through a configuration file, keep your API credentials secure and never commit them to source control.
Apptrove MCP server address: https://mcp.apptrove.com/mcp
The exact configuration steps may vary by AI client, but the underlying connection provides access to Apptrove reporting through the same MCP server.
Security and Access Controls: Read-Only by Design
Connecting an AI assistant to attribution data raises an important question: how do you make reporting easier without giving the assistant unrestricted access to your account?
Apptrove MCP is designed around read-only access, explicit request approval, and application-level permissions.
Read-Only Access
Apptrove MCP can retrieve information, but it cannot create, modify, or delete anything in your Apptrove account.
This allows you to explore reporting data through an AI assistant without giving it permission to change your account’s data.
Approval Before Execution
Every request the assistant makes through Apptrove MCP is presented to you before execution.
You can review the request and approve or decline it, maintaining control over which requests run.
Permission-Scoped Access
Access through Apptrove MCP mirrors the permissions associated with your Apptrove account and API key.
- Advertisers can access their own applications.
- Agencies can access the applications they have been granted permission to manage.
- Partners can access their own traffic.
- Administrators and members within an organisation can access the applications assigned to them.
API keys are scoped to specific applications, and the assistant cannot access applications outside that scope.
Tenant Isolation
Apptrove MCP maintains isolation between customers. One customer’s data cannot be accessed through another customer’s connection, and this has been verified through testing.
Rate Limiting and Prompt-Injection Testing
Request limits help protect accounts against abuse if an API key is compromised.
The server has also been tested against prompt-injection attempts involving text embedded in data fields, such as partner and campaign names. These safeguards are intended to reduce the risk of malicious content manipulating the assistant’s behaviour.
Together, these controls help teams use AI-assisted reporting while retaining defined access boundaries and control over data retrieval.
How Teams Can Use Apptrove MCP
Apptrove MCP can support everyday reporting as well as more specialised investigations across marketing, analytics, and technical operations.
Prepare the Morning Performance Brief
Instead of manually assembling a daily report, a growth manager can ask for yesterday’s installs, events, and revenue by partner, with the largest movements highlighted.
The result provides a starting point for the morning stand-up, helping the team identify which changes need attention.
Build Funnels Without Waiting for Custom Reports
Consider an advertiser that needs a monthly view of acquisition by source, tracking the journey from lead to account opening to first trade.
Rather than waiting for a bespoke dashboard, the team can ask its connected AI assistant to retrieve the relevant funnel data and conversion ratios through Apptrove MCP.
Review Partner Quality Before Monthly Meetings
Before a partner review, a user can ask which partners generated the most rejected installs, the reasons for rejection, and how post-install engagement compares across sources.
This brings traffic quality and engagement into the same investigation, helping teams prepare for more informed partner discussions.
Troubleshoot Missing Conversions
When a partner reports missing conversions, a support or integration engineer can request postback and event log samples for the relevant period.
The assistant can summarise what was sent, received, and rejected, helping the team determine where further investigation is needed.
Compare Results Across Multiple Applications
Agencies and advertisers managing several apps can compare performance across their accessible applications in a single question.
This reduces the need to repeat the same reporting process for each app and makes cross-app analysis more convenient.
Who Is Apptrove MCP For?
Apptrove MCP is designed for teams that regularly work with mobile attribution data and need a more direct way to retrieve and interpret reporting information.
- User acquisition and growth managers can review campaign performance, identify changes in acquisition metrics, and investigate partner quality.
- Performance marketers can compare channels, partners, and campaigns without manually assembling every report.
- Agencies can analyse performance across the applications they have permission to access.
- Data analysts can retrieve reporting data more quickly before carrying out deeper analysis or modelling.
- Support and engineering teams can examine event and postback logs when investigating integration issues.
Whether you work primarily with campaign reporting, cohort analysis, or technical troubleshooting, Apptrove MCP gives you another way to access the information already available in your Apptrove account.
Get Started With Apptrove MCP
Apptrove MCP is available to Apptrove customers. Getting started involves connecting a supported AI assistant to your Apptrove account.
Follow these steps:
- Generate an Apptrove API key. Create a key for the applications you want your assistant to access, with the appropriate permissions.
- Add the MCP server. Configure Apptrove MCP as a custom connector in your AI assistant using https://mcp.apptrove.com/mcp.
- Authenticate your connection. Follow your AI client’s authentication prompts and provide your Apptrove credentials as required.
- Ask your first question. Start with a broad request, such as “Summarise last week’s performance across my apps,” and refine your questions as needed.
Apptrove MCP supports MCP-compatible clients, including Claude, ChatGPT, Cursor, and Claude Code. Configuration details can vary depending on the client you use.
If you need help setting up the connection, contact your Apptrove account manager.
Your attribution data is already telling a story. With Apptrove MCP, you can spend less time searching for the numbers and more time understanding what they mean.
Connect Apptrove to your AI workflow and start exploring your attribution data through natural-language questions.