Geckoboard Metrics MCP

Accurate metrics
for your AI assistant.

Give your AI access to real-time metrics from tools like HubSpot, Aircall, and Google Analytics. Use it to analyze changes, automate workflows, and build your own tools on data you can trust.

Start free trialSchedule a demo
  • Read-only access
  • Token efficient

Trusted by thousands of data-driven teams

ClaudeFileEditWindow9:41
What would you like to know?

Do more with your data

Use the AI tools of your choosing, confident your metrics are accurate. Add context and combine data from multiple sources to go further.

Quickly pull back answers
Need to know how many tickets you received last week regarding billing? Simply ask and get the answer. No wrestling with in-app reports.
Kick off open-ended investigations
Unsure why sales seem to have been slow last week? Kick off a thorough investigation, receive an answer, and ask follow-ups. What would have taken days now takes minutes.
Create reports that tell a story
Replace manual reporting with custom Claude and ChatGPT artifacts, complete with narratives. Use them for coaching, client reporting, and weekly meetings, or to replace tired spreadsheets.
Automate audits
Routinely performing audits? Create skills you can reuse to do anything from reviewing clients against best practices, through to reviewing your own SLA adherence.
Vibe code your own dashboards
Need a hyper-bespoke dashboard? Use tools like Lovable, Claude Code, or ChatGPT’s Sites to build your own custom dashboard on top of data you can trust.
Combine data from multiple sources
Headless access to your data allows you to combine data across multiple tools, mix qualitative and quantitative data, and benefit from the context your assistant has already built up around your business.
Start free trialSchedule a demo
quotation mark icon

The MCP server is a brilliant added value to my team and my overall analysis, it’s great for annual reports and trends. I’ve started to use it on a daily basis.

A picture of Orad Vinograd
Orad Vinograd
Support Director at Driivz
quotation mark icon

It’s like having a data analyst without having a data analyst.

A picture of Zach Gilbert
Zach Gilbert
Director Customer Operations at COBRA Firing Systems

AI needs pre-computed metrics

AI assistants like Claude, Gemini, and ChatGPT make excellent analysts, but only when they have access to pre-computed metrics.

Working from raw records

Most MCPs are designed for returning and performing actions on records. Your AI assistant cannot simply ask a question and get an answer.

Instead it has to use brute force: first figuring out the business logic, then pulling back all the records and deciding what fields map to what, before applying a script to perform the calculation.

The result?

  • Missed records
  • Guessed business logic
  • Inconsistent time periods
  • Slow answers
  • Heavy API usage
  • Heavy token usage
  • Limited to basic analysis
Claude asks for the first response time, then has to research and define first response time, decide a strategy, fetch raw tickets, and run a script to calculate the metric — hitting the Zendesk API multiple times — before giving an incorrect answer of 5m 13s
Claude asks Geckoboard MCP for the first response time, which requests the metric from Geckoboard’s database and gives the correct answer of 4m 20s

Getting metrics directly

With the Geckoboard Metrics MCP, your AI agent can simply ask for the KPIs it needs, filtered and segmented as necessary, and the MCP will return the answer.

We’ve already done the work to transform the raw data into KPIs.

The result?

  • Accurate metrics
  • Complex analysis possible
  • Lower token usage
  • No API limit concerns
  • Faster answers

Keep source systems and sensitive data protected

A team reviewing a Geckoboard dashboard together on laptops

IT teams are rightly cautious about giving AI assistants direct access to systems of record. With write access, there’s always a risk that data could be changed or corrupted.

Geckoboard provides a layer of separation, giving your AI assistant access to the metrics it needs without write access to your source systems. Our data-source connections are read-only, as is the Metrics MCP’s access to data held in Geckoboard.

We also minimize PII by retrieving only the data needed to calculate your metrics, not the contents of tickets, calls or conversations.

Getting started

You can be up and running with the Metrics MCP in less than 30 minutes.

An AI chat asking why support tickets dropped, alongside the logos of connected data sources like Zendesk, HubSpot, and Shopify
  • 1

    Create a Geckoboard account

    Start a 14-day free trial. The MCP is available on all plans.

  • 2

    Connect your data sources to Geckoboard

    Connect your data sources to Geckoboard to access their metrics through the MCP.

  • 3

    Connect Geckoboard to your AI assistant

    Follow our setup instructions to connect the Geckoboard Metrics MCP to your preferred AI assistant or tool.

Supported integrations

Currently the Metrics MCP supports over 15 integrations, and we’re adding more.
Have a particular data source you need us to provide the semantic layer for? Get in touch.

hubspot logo
zendesk logo
aircall logo
google-analytics logo
pipedrive logo
facebook logo
google-ads logo
google-search logo
gorgias logo
intercom logo
linkedin logo
shopify logo
See all MCP-enabled integrationsArrow icon
Start free trialSchedule a demo
FAQ

The Geckoboard Metrics MCP is designed for everyone who works with data as part of their role, from senior leaders through to frontline agents, analysts and ops roles.

Combined with your AI assistant, the MCP enables anyone to perform complex analysis quickly using trustworthy metrics.

Compared with traditional reporting, it significantly reduces the barrier to entry, allowing anyone to interact with their data conversationally.

Use it to ask questions like “Why did the UK see a big drop-off in new deals created last week?” through to “Was there a link between Shopify orders last month and support tickets created?”

The only requirements are an interest in data, access to an AI assistant that supports MCP and a Geckoboard Editor seat.

No. The Geckoboard Metrics MCP can request any metric, with any filter, breakdown or timespan that Geckoboard supports. You are not limited to what you display on your dashboards, and you do not need to configure anything upfront.

The Geckoboard Metrics MCP currently supports metrics from Aircall, Dialpad, Facebook, Google Analytics 4, Google Ads, Google Search Console, Gorgias, HubSpot, Instagram, Intercom, LinkedIn, Matomo, Pipedrive, Shopify, Stripe, YouTube and Zendesk, with more being added all the time.

You can connect multiple accounts from the same data source and work with multiple data sources at once.

A full list is maintained here.

If you need an integration that isn’t there yet, tell us.

The Geckoboard Metrics MCP works with any service that supports MCP. That includes assistants like Claude and ChatGPT, build tools like Lovable and Replit, and automation platforms like Zapier and n8n.

As more tools adopt the standard, we’ll continue to extend our set-up instructions.

Most data sources do not provide metrics via their APIs. This means your AI assistant cannot just ask for a metric and get the answer.

Instead, it will try its best via brute force. It pulls back raw data and tries to calculate the metric itself. This is error-prone, slow and expensive in terms of AI tokens. It also usually involves giving your AI assistant write access to your data source, as well as access to sensitive customer data.

Furthermore, many services haven’t yet released even a basic MCP server.

No. The Geckoboard Metrics MCP is read-only. Geckoboard also only uses read-only connections to access your raw data. This means it cannot change the data in your source systems.

The Geckoboard Metrics MCP requires a Geckoboard account. You can start a 14-day free trial here.

You then need to connect Geckoboard to your data sources and follow our instructions for connecting the Geckoboard Metrics MCP to your AI assistant. The whole process can be done in a few minutes.

To connect the Geckoboard Metrics MCP to your AI client, you must currently be a Geckoboard Editor.

Tools like Claude and ChatGPT require each user of the MCP to connect Geckoboard separately.

The permissions required to connect your data source to Geckoboard vary by service. Geckoboard only requires read-only API access.

The MCP is available to try for free as part of the Geckoboard 14-day free trial.

It is currently included with all Geckoboard plans. There is no usage-based charge for the MCP. Each user of the MCP will likely require their own Geckoboard Editor seat.

You only consume tokens in your AI client, so you don’t need to worry about incurring additional Geckoboard charges for using the MCP.

No. The Metrics MCP is currently part of our combined offering, which includes dashboards and alerts. All Geckoboard customers have access to it.

They do different jobs. The purpose of Geckoboard’s live dashboards is to give teams a shared, real-time view of their current performance. Dashboards are still the best way of achieving this.

The MCP replaces analysis and ad hoc reports, and enables you to build custom tooling.

No, not yet. We are working on a separate MCP server for working with your dashboards.