Quantum Metric

There was a ton of healthy pressure earlier in the year for everyone to answer the question “what’s our MCP strategy.” It’s a valid question, but MCP is a protocol. For me it’s like asking what your HTTP strategy is... well, we have a web-based application you can load in a browser. So what? To be fair, I appreciate the significance of the MCP question given the rise of AI. Early on, a lot of this was wrapping pre-AI-age APIs in some context and shipping it to give agents the ability to do some CRUD-ing! A fun, and good first step. But we wanted to think through what it would take for an agent we don’t control to use our data to take high-impact action. I’m always looking for a good analogy. For me, this is very much like providing a kitchen to a chef. We don’t get to hire the chef, but we do get to decide what’s in the kitchen. What I wanted to make sure we absolutely avoided was making a lot of noise about our kitchen, only to find out it was a freezer, a microwave, and a dozen TV dinners. No shade, TV dinners just helped me make it through a chaotic week of travel. But what’s the point of having a chef then? What if instead, we gave the chef the equipment, utensils and ingredients to “let them cook?’ (sorry, had to). And the cookbooks and guidance to help them use what’s there, without prescribing every meal they can make. Think raw query-building against real-time experience data, with all the context and guides an agent needs to do it well, data discovery tooling for the agent to find its own creative solutions, and intelligence feeds that take extremely sophisticated pipelines to generate, delivered so an agent can jump straight to taking action. That’s how we approached our MCP, which we released this week. Give agents the tools, intelligence feeds and knowledge to do better analysis with Quantum Metric. And give them room to do things with those tools that we haven’t thought of yet.

Quantum Metric MCP is generally available today, and the early customers using it don't use it the way they would use our UI. They don't ask for a dashboard or a funnel. They ask what they actually need to know. Which problems are hurting customers most right now? What changed after the last release? The agent recognizes that Quantum Metric has the data to answer, so it pulls it in. Most of these teams already had agents connected to their monitoring tools. Adding our MCP took them a few minutes, and now their agents can see what customers went through, not only what broke on the back end. If you're a Quantum Metric customer, you can connect it today. The full announcement is in the comments.

Over the past decade, I've watched digital teams uncover countless issues, each with the potential to drive millions of dollars into their brand's bottom line. But that potential often hasn't translated into action. And I've understood why, because I've lived in Product for 20+ years myself. Action has required so much more than just a find: handoff to the right party, ticket creation, prioritization, sprint planning, development, testing, staging and release. We've all been busy jumping through a bunch of necessary hoops. I'm excited to watch technology advancement strip away those layers of process and focus our jobs down to what they should be: building something that makes our customers happier, every day. For the past year at Quantum Metric, we've been using the MCP that we're now releasing to our customers, bringing the experience context of our own QM users directly into the AI coding harnesses that our Engineers steer. This has refocused our dev work onto optimizations that have the greatest impact on our customers. We understand each experience using data, instead of hearsay. We directly quantify customer impacts and cohorts before over-rotating on solutions. We act surgically on things that actually impacted the customer, with faster, more effective bug fixes. And we measure the impact of our production changes, directly from the AI coding tools where we just made those changes. I can't wait to watch our customers do this for themselves. Let's go!

More than a decade of building Quantum Metric and working with digital enterprise has taught me a painful lesson: 𝐭𝐚𝐥𝐞𝐧𝐭𝐞𝐝 𝐩𝐞𝐨𝐩𝐥𝐞 𝐜𝐚𝐧 𝐰𝐨𝐫𝐤 𝐢𝐧𝐜𝐫𝐞𝐝𝐢𝐛𝐥𝐲 𝐡𝐚𝐫𝐝 𝐨𝐧 𝐭𝐡𝐞 𝒘𝒓𝒐𝒏𝒈 𝒕𝒉𝒊𝒏𝒈𝒔. A big part of 𝐥𝐞𝐚𝐝𝐞𝐫𝐬𝐡𝐢𝐩 is helping a team figure out what deserves its attention. That means knowing your customers, making tradeoffs, and asking whether the work is actually moving the business forward. 𝐀𝐈 𝐀𝐠𝐞𝐧𝐭𝐬 𝐡𝐚𝐯𝐞 𝐭𝐡𝐞 𝐬𝐚𝐦𝐞 𝐩𝐫𝐨𝐛𝐥𝐞𝐦. We keep giving them more capability, more autonomy and more work. But, how do they know what’s worth doing? A ticket tells an agent what someone asked for. An error log tells it something broke. Neither one tells it what customers actually experienced, how many were affected, what it cost the business or if the fix actually made things better for the customer. That’s the context an agent needs to decide what matters. It’s what we call 𝐄𝐱𝐩𝐞𝐫𝐢𝐞𝐧𝐜𝐞 𝐂𝐨𝐧𝐭𝐞𝐱𝐭. Today we’re making 𝐐𝐮𝐚𝐧𝐭𝐮𝐦 𝐌𝐞𝐭𝐫𝐢𝐜 𝐌𝐂𝐏 generally available, so the agents enterprises already use can tap into that context. I’ve cared about helping teams maniacally focus on what matters for a long time (their customers). Hearing how customers are putting this into practice makes this launch especially meaningful to me (as Dylan illustrates below). As agents take on more of the work, leaders have to get even better at giving them direction.

Every enterprise is building agents. Almost none of them can see what customers actually experienced. Monitoring tells an agent something broke, but it can’t tell it which customers felt it or what it cost the business. So agents burn tokens fixing the wrong things: quickly, confidently, but still wrong. Give your agents better judgment with Quantum Metric's experience context via MCP. Connect once and any MCP-compatible agent gets 46 tools built for agents, backed by the richest first-party experience data set in the enterprise. Here’s what teams are already doing with it: 1. Self-improving websites: coding agents detect, fix and measure friction in a continuous loop 2. Incident triage ranked by customer impact. Royal Caribbean Group is using it to catch bugs that would have slipped through triage and hand developers the hard numbers to act. 3. Test roadmaps prioritized by what actually moves the business. Optimizely is now live in the directory, so their AI knows which experiments matter most before anyone hits launch. As agents take on more of the work, leaders have to get even better at giving them direction. Can't wait to see what you build with it :)

Every analytics vendor is shipping an MCP server right now. We’re already seeing “support MCP” show up on RFPs, and if it isn’t on yours yet, it should be. Trouble is, it means about as much as “we have an API,” because an MCP server is just a doorway. What matters is what your agent finds when it walks through. Here’s the thing most of them do: wrap the reporting API they built a decade ago and call it agent support. It demos great. Then three weeks in, someone asks the question nobody pre-built a report for, and the agent just stalls. New protocol, decade-old ceiling. Quantum Metric went a different way, partly by accident: We never shipped a public reporting API, so when we built our MCP server there was nothing old to wrap. We got to design it for agents from scratch, pointed at the full experience data we capture. That means an agent can compose its own queries across the model, drill into session replay, and discover what exists on its own, instead of picking from a fixed menu of yesterday’s reports. My rule for anyone evaluating an “agentic” analytics tool: bring it a genuinely novel question, one no one would’ve built an endpoint for, and watch whether it investigates or freezes. That single test tells you which kind of server you’re dealing with. More on this in my latest post: https://lnkd.in/guBDZbXX