Your analyst, inside Claude, ChatGPT and Cursor.
ConvoData runs as an MCP server. Add it once and the same cross-source answers you get in ConvoData show up wherever you already work.
Where it works
Skip the tab. Ask from anywhere.
Claude and ChatGPT
Add the connector, sign in once, and ask across every connected source without leaving the conversation.
Cursor and the terminal
Wire it into Cursor or Claude Code and script the audits you run every week.
Slack
Tag the workspace bot in any channel and the answer posts as a thread your team can read.
One server URL. That's it.
Add ConvoData as a remote MCP server in your client, sign in, and your connected sources are ready to query.
- Read-only by default
- Uses the sources already connected in your workspace
- Same metric definitions as the ConvoData app
The difference
Single connectors vs. ConvoData MCP.
| What you need | One connector per platform | ConvoData MCP |
|---|---|---|
| Questions across several sources | Each platform read on its own | All sources joined first |
| Consistent numbers | No shared metric definitions | Same answer every time |
| Checking the maths | Hard to trace | Answers show their sources |
| Setup | One connector per tool | One server URL |
FAQ
Common questions.
MCP (Model Context Protocol) is an open standard that lets AI assistants like Claude and ChatGPT use outside tools and data. ConvoData's MCP server gives them access to your connected marketing data.
A single connector reads one platform in isolation. ConvoData joins your sources into one layer first, so a question spanning organic, paid and revenue returns the same answer every time.
No. The MCP server is read-only by default, so it can answer questions but not edit your campaigns or settings.
No. Your data is queried on demand, encrypted in transit and never used to train models.
Connect · Analyze · Act
Explore the rest of ConvoData.
Bring your data into the AI tools you already use.
Connect your sources once, then ask from Claude, ChatGPT, Cursor or the terminal.