Skip to main content

Geckoboard MCP: Frequently Asked Questions

Geckoboard MCP lets a compatible AI assistant request metrics from supported data sources connected to your Geckoboard account.

Before you start:

  • MCP access is currently included with every Geckoboard plan. There is no separate Geckoboard MCP fee.

  • The Geckoboard user who authorizes the connection must be an editor.

  • You also need an account with an AI tool that supports MCP. That provider may have its own plan or usage charges.

What is MCP?

MCP stands for Model Context Protocol. It is a standard that allows an AI tool to request information from another service.

Geckoboard runs an MCP server that gives compatible AI tools read-only access to metrics calculated by Geckoboard. The AI tool uses those results to answer your question or complete a task.

Geckoboard MCP is not an AI assistant itself. The wording, interpretation and any additional calculations in the final answer come from the AI tool you use.

What can I use Geckoboard MCP for?

You can use it to:

  • ask for a metric without opening a dashboard;

  • compare a metric across time periods;

  • break down a metric by an available field, such as team, agent or channel;

  • investigate a change by requesting related metrics and breakdowns;

  • calculate a ratio or other value from metrics returned by Geckoboard;

  • supply metrics to reports, workflows or tools built in an MCP-compatible service.

The available metrics, filters, time periods and breakdowns depend on the data source.

How is this different from connecting an AI tool directly to a data source?

A direct connection may give an AI tool access to individual records, such as tickets, calls or deals. The assistant then has to work out how to calculate a metric from those records.

Geckoboard MCP provides metrics calculated by Geckoboard, including rates, durations, medians and period comparisons where supported. It is intended for metric questions rather than general record management.

You may use both types of connection if you need calculated metrics as well as record lookup or actions in the source system.

What do I need before I connect?

You need:

  1. A Geckoboard account and an editor role on that account.

  2. At least one supported data source connected to Geckoboard.

  3. An AI tool and plan that support MCP connections.

Connecting the MCP does not connect Zendesk, Aircall or another source to Geckoboard. Set up the data-source connection first and allow any initial data import to finish.

How much does Geckoboard MCP cost?

Geckoboard MCP is currently included with all Geckoboard plans. Geckoboard does not charge a separate MCP fee.

The person who connects Geckoboard to an AI tool must be a Geckoboard editor. Your plan’s normal editor allowance and any charge for additional editors still apply.

Your AI provider’s subscription and usage charges are separate from your Geckoboard subscription. The data-volume allowance on your Geckoboard plan also continues to apply.

Who can connect Geckoboard to an AI tool?

The Geckoboard user who completes the authorization must be an editor. View-only users cannot authorize an MCP connection.

The connection is approved through Geckoboard’s sign-in and authorization flow. It does not require you to give your Geckoboard password to the AI tool.

Which AI tools can I use?

You can use Geckoboard MCP with an AI tool that supports remote MCP servers and Geckoboard’s authorization flow.

  • Claude: Geckoboard is available in Anthropic’s Connectors Directory.

  • ChatGPT: Full MCP support is currently in beta and is not available on every ChatGPT plan. Setup uses Developer Mode and a custom connector.

  • Other tools: Use the custom MCP server option if the tool supports HTTP or Streamable HTTP transport and browser-based authentication.

Availability and setup options are controlled partly by the AI provider and may change independently of Geckoboard.

How do I connect Geckoboard to Claude?

  1. Open Claude on the web or desktop.

  2. Go to Customize.

  3. Select Connectors.

  4. Select Add, then Browse connectors.

  5. Search for Geckoboard.

  6. Open the Geckoboard result and select Add.

  7. Sign in to Geckoboard and authorize the connection when prompted.

How do I connect Geckoboard to ChatGPT?

Full MCP support in ChatGPT is currently in beta and may not be included with your ChatGPT plan.

  1. Enable Developer Mode in ChatGPT on the web.

  2. Go to Plugins.

  3. Select the + icon next to the search box.

  4. Enter these details:

    • Name: Geckoboard

    • Connection: Server URL

    • URL: https://gateway.geckoboard.com/mcp

  5. Accept ChatGPT’s warning about Developer Mode.

  6. Select Create.

  7. Select Sign in with Geckoboard and complete the authorization flow.

How do I connect another MCP-compatible tool?

The labels vary between tools, but the usual process is:

  1. Open the AI tool’s settings.

  2. Find the option to add a custom MCP server.

  3. Enter https://gateway.geckoboard.com/mcp as the server URL.

  4. Select HTTP or Streamable HTTP if the tool asks for a transport.

  5. Sign in to Geckoboard and authorize the connection.

  6. Save and enable the connection in the AI tool.

Refer to the AI tool’s documentation if it does not offer these options.

Which Geckoboard data sources does the MCP support?

The current setup guide lists:

  • Aircall

  • Facebook and Facebook Ads

  • Google Ads, Google Analytics and Google Search Console

  • HubSpot

  • Instagram

  • Intercom

  • LinkedIn and LinkedIn Ads

  • Pipedrive

  • Shopify

  • Stripe

  • YouTube

  • Zendesk

This is a subset of all data sources available in Geckoboard. The MCP can only query a supported source after that source has been connected to your Geckoboard account.

Does it use the same metrics as my dashboards?

The MCP and Geckoboard dashboards use metrics from the same Geckoboard metrics engine. The MCP lets the AI tool request those metrics directly; it does not read the visual layout of a dashboard.

When comparing an MCP result with a widget, make sure both use the same connection, metric, timeframe, timezone, filters, aggregation and breakdown.

Can the MCP change my data or dashboards?

No. Geckoboard MCP is read-only. It cannot edit source records, tickets, dashboards or account settings.

If an AI tool performs another action after querying a metric, that action comes from a different connection or capability in the AI tool, not from Geckoboard MCP.

Can it automate a report or alert?

The MCP supplies metric data when an AI tool requests it. Scheduling, document creation and message delivery depend on the features and permissions of the AI or workflow tool you use.

Connecting Geckoboard MCP alone does not create a scheduled report, monitor a metric continuously or send notifications.

Are there limits on the results it returns?

Yes, although most metric questions return small results and will not reach them. Limits on record lists and category breakdowns apply to what is returned to the AI tool, not to the data Geckoboard uses to calculate the metric.

See Understanding result sizes in Geckoboard MCP for the current limits and ways to narrow a large request.

How should I phrase a question?

Include the metric, timeframe and any filters or breakdowns you need. For example:

Use Geckoboard MCP to compare Zendesk first reply time this week with the previous four-week average. Break the result down by group.

Use Geckoboard MCP to show Aircall waiting time by team for the past seven days and identify the largest change from the previous seven days.

Ask the assistant to state the metric, filters and time period it used if you need to check the result against Geckoboard.

Related Articles
Did this answer your question?