Heatmaps Don't Work in Three Dimensions: Rethinking Analytics for Spatial Web Experiences

Let's be honest for a second. Most developers building spatial web experiences right now are still pasting a Google Analytics snippet into their project and calling it a day. Maybe they're logging a custom event when a user "enters" a 3D scene, or firing a conversion pixel when someone completes a gesture-based interaction. And technically, sure, that's something. But it's a little like measuring a symphony by counting how many notes were played.

The spatial web isn't a flat surface. So why are we still measuring it like one?

The Fundamental Mismatch Between Old Metrics and New Experiences

Traditional web analytics were built around a core assumption: users move through a two-dimensional plane. They scroll down. They click left or right. They spend time on a page before bouncing to another. Every metric we've relied on for the past two decades — session duration, bounce rate, click-through rate, scroll depth — is rooted in that flat-page mental model.

Spatial web experiences break every single one of those assumptions simultaneously.

When a user is navigating a volumetric product configurator, they might spend 45 seconds rotating a 3D model without ever "clicking" anything in the traditional sense. They might physically lean forward, use a hand gesture to scale an object, look away from a UI element that's technically in their viewport, or dwell on a specific corner of a 3D environment for reasons that have nothing to do with intent. None of that shows up in your current dashboard. None of it.

The problem isn't just that your tools are missing data. It's that the data they are collecting is actively misleading you.

What Spatial Engagement Actually Looks Like

Before you can measure something, you need to know what you're measuring. And in 3D environments, "engagement" takes on a whole new set of dimensions — literally.

Gaze and attention tracking is probably the most immediately useful signal in spatial experiences. Where is the user looking, and for how long? This isn't the same as tracking cursor hover states. Spatial attention is dynamic, it shifts in three axes, and it tells you things about cognitive load and interest that click data never could. Tools like the WebXR Device API expose gaze vector data when users are in XR sessions, giving developers a raw stream of directional attention that can be mapped against scene geometry.

Gesture pattern analysis is another layer entirely. Are users completing gestures fluidly, or are they fumbling and retrying? A high gesture retry rate on a specific UI element is the spatial equivalent of a rage click — it's a friction signal screaming at you that something is wrong. Logging gesture velocity, completion rate, and error recovery patterns gives you a picture of UX health that no bounce rate ever could.

Spatial dwell mapping — essentially a 3D heatmap of where users spend time in a volumetric space — is arguably the most powerful tool in this new toolkit. Think of it as scroll depth's smarter, more spatially aware cousin. Instead of tracking how far down a page someone scrolled, you're mapping which regions of a 3D environment attracted sustained attention. Several emerging frameworks, including some built on top of Three.js and Babylon.js, are starting to bake this kind of instrumentation directly into their scene graph logic.

The Tools That Are Actually Emerging

The honest answer is that the spatial analytics ecosystem is still pretty early. There's no Mixpanel for XR yet — not really. But a few players are starting to stake out territory worth watching.

Cognitive3D has been one of the more serious efforts to build analytics infrastructure specifically for 3D and XR experiences. Their platform captures gaze data, positional heatmaps, and session replays in volumetric space — which is genuinely wild to see in action. It's been used more heavily in enterprise VR training contexts, but the underlying architecture applies directly to spatial web development.

Unity Analytics and Unreal Insights are relevant if your spatial web experience has roots in game engine tooling, which a growing number do given the WebAssembly pipeline options available today. The session replay and event funnel capabilities there are more mature than anything web-native right now.

For developers working closer to the browser layer, the WebXR Analytics Proposal — still a draft discussion in W3C circles — hints at a future where browser vendors expose standardized spatial telemetry hooks. That's not here yet, but keeping an eye on that thread is worthwhile.

In the meantime, a lot of teams are rolling their own solutions: custom event schemas that capture position vectors and rotation quaternions alongside traditional event metadata, piped into whatever data warehouse they're already using. It's more work, but it gives you ownership of the schema from day one.

Rethinking Your Data Collection Strategy from the Ground Up

If you're building a spatial experience right now and you want to do this right, here's the mindset shift that matters most: stop thinking about events and start thinking about states.

Traditional analytics are event-driven. Something happens, you log it. Spatial analytics need to be state-aware. Where is the user in 3D space? What are they looking at? What's their body posture relative to the experience? These aren't discrete events — they're continuous streams of context that you sample at regular intervals and make sense of in aggregate.

Practically, this means:

• Define your spatial zones before you write a single analytics call. Divide your 3D environment into meaningful regions and assign semantic labels to them. "Product shelf," "info panel," "exit zone" — whatever makes sense for your experience. This is your spatial taxonomy, and it's the foundation everything else builds on.
• Log rotation and orientation, not just position. A user standing in front of a product display facing away from it is a very different signal than one facing toward it. XYZ coordinates alone miss that completely.
• Sample continuously, not just on interaction. Set up a positional telemetry loop that fires every 500ms or so during active sessions. Yes, it's more data. Yes, it's worth it.
• Build in gesture quality scoring. Don't just log that a gesture fired. Log whether it completed cleanly, how many attempts it took, and how long it took from initiation to completion.

The Privacy Layer You Can't Ignore

One thing that doesn't come up enough in these conversations: spatial data is deeply personal in a way that page view data isn't. Gaze tracking, positional data, and gesture patterns can reveal things about users' physical abilities, cognitive patterns, and even emotional states. That's not hypothetical — researchers have demonstrated that gaze patterns alone can indicate neurological conditions.

Before you instrument your spatial experience with this kind of telemetry, make sure your privacy policy is updated, your consent flows are explicit, and you've thought hard about data minimization. The FTC has been increasingly aggressive about data practices that users don't expect, and spatial telemetry is exactly the kind of thing that could land you in a very uncomfortable spotlight.

Start Measuring Like the Web You're Actually Building

The spatial web is real, it's here, and it's only going to get more immersive as WebXR matures and device adoption accelerates. The developers who figure out how to actually measure what's happening in these environments — not just approximate it with legacy tools — are going to have a massive competitive advantage.

Flat analytics for a three-dimensional web is like trying to understand a city by only looking at its map. You're getting the shape, but you're missing everything that makes it alive. Time to build better instruments.