Chatting With Your Analytics Data Is Now a Thing - Make Sure It's Telling the Truth

So, what does that mean? In short, you can now chat with your GA4 data using natural language. That's right, no more complex report building, no more digging through endless menus. Just ask a question, and Google's AI will generate an answer based on your analytics data; that's pretty neat, right?

Here's the official announcement from Google: Model Context Protocol (MCP). The short launch video is well worth watching.

You can also explore the codebase on GitHub: MCP GitHub Repository, and join the growing conversation about the innovation on Discord: Join the Discord Community.

We're already working on building interfaces to take full advantage of this. It opens up exciting new possibilities for marketers, analysts, and decision-makers to access meaningful insights instantly. And, if it works as well as the potential suggests, then marketers who have struggled to get to grips with analysing data on the platform will greatly enjoy the innovation.

But while we're excited (and we truly are), this innovation also comes with a serious word of caution.

The introduction of the MCP server by the Google engineering team introduces a fundamental shift in how we interact with analytics. For example, you could now ask a question like:

"I have a monthly budget of £5k, tell me how to optimally spend this budget to achieve the biggest possible revenue uplift."

You'll receive a response based on your actual GA4 data; however, whether the resulting recommendation is effective (or not) is the $64,000 question. 

Either way, for now, this leap forward means AI can automate many of the repetitive, time-consuming, but business-critical tasks we do every day:

• Running analytics audits
• Performing data health checks
• Monitoring conversion events and traffic anomalies
• Recommending optimisation actions

We're already training AI agents to handle these tasks using the MCP interface. It's early days, but we expect polished, user-friendly interfaces to roll out industry-wide within the next month or two.

Now here's the part we need to stress: AI doesn't know when (or if) your data is bad. And that's the real risk.

The same AI that can give you marketing recommendations based on your data will also confidently suggest the wrong things if your data quality is poor.

It won't flag inconsistencies, broken tracking, untagged campaigns, or gaps in conversion attribution. It will just answer.

If you haven't been regularly auditing your data, ensuring consistent tracking, and fixing issues as they arise, this new capability might do more harm than good.

Because AI doesn't second-guess your data, you need to be sure you can trust it before you ask it to make decisions.

Google has built the MCP on top of the existing Google Analytics Data API, which means it inherits some of its limitations:

• API quotas: The default limit is 50,000 tokens per day per property. Heavy use of chat-based querying could hit this quickly.
• Data sampling and thresholding: Like standard GA4 reports, your results may be subject to data sampling or thresholding, depending on your settings and traffic volume.
• Data retention limits: GA4 data retention defaults to two months (which you can extend to 14 months) unless configured otherwise.

All of this underscores the need for smart setup and proactive maintenance. You want your data structured and stable enough to handle these new tools responsibly.  

We're treating this announcement as a green light to accelerate our work on AI agents for data monitoring and optimisation. But importantly, we're doubling down on what we already do for our clients every month:

• Deep GA4 data audits
• Proactive data quality health checks
• Fixes for broken or misconfigured tracking
• Continuous optimisation of reporting setups

Because the only thing more exciting than asking your data thoughtful questions via chat is getting helpful, incisive and trustworthy answers in return.

The introduction of the MCP server is a turning point. It marks the beginning of a new era in analytics where natural language meets real-time data.

But it also puts your data's quality under a spotlight. If it's not clean, complete, and well-structured, the insights you get from AI won't just be useless; they might be misleading. That said, it should not stop you from having a play with the shiny new toy.

Before experimenting with the new feature, do keep in the back of your mind that the quality answers you seek to power performance will only be as good as the quality of data going in.

The future is here. However, don't forget that it requires clean data to function correctly.

We offer one-time audits and ongoing monthly support to make sure your Google Analytics setup is fully optimised, quality-checked, and ready for the AI-powered future. Learn more about our auditing process.

If you want to check your data integrity yourself, you can find our complete health check process documented and free to download. 

Let's get your data AI ready. Reach out today.