Sizzle vs. Substance: What We Actually Reward in IoT Dashboards

By Damian Bucovsky, President at The Shadow on the Moon

Imagine you're sitting in a demo of a monitoring platform for an industrial facility. The centerpiece is a 3D rendering of the plant. You can fly the camera over the roofline, swoop down between tanks and pipes, and click on a piece of equipment to see... a single sensor reading. Temperature, maybe a pressure value, a green or red dot.

It's genuinely impressive to watch. The executives in the room love it. It looks like something out of a video game, and it makes the company procuring it feel like they're buying the future. And it is, in almost every way that matters to the people who'd actually use it every day, worse than a table.

Two very different audiences, two very different definitions of "good"

Here's the thing that's easy to lose sight of: a dashboard has at least two audiences, and they are judging it by completely different criteria.

The buyer, often an executive, a procurement committee, someone signing off on a budget line sees the dashboard a handful of times. Maybe in a sales demo, maybe in a quarterly review, maybe when showing it off to a visiting client. And it's rarely just the buyer in the room: it's their customers, their investors, the board. For that audience, the 3D flyover works, and it works for a real reason, not an irrational one. It's memorable. It signals investment and sophistication. It's a tangible thing to point to in an investor update or a client tour. That's not nothing; it's real value, just not the operational kind. It's value in perception, in confidence, in the story a company can tell about itself. And it's easy to justify a bigger number on the invoice when the thing you're buying looks like it belongs in a control room from a movie.

The operator, the person who has to open this thing at 6am, again at noon, again before they go home, every single day for years has a completely different relationship with it. They're not there to be impressed. They're there to answer a specific, boring, high-stakes question as fast as possible: is anything wrong, is anything trending toward wrong, and what do I need to do about it right now. For that job, a compound view of historical trends, thresholds, and deviations is worth infinitely more than a rotating 3D model of a building they walk through every day and don't need a virtual replica of.

The flyover isn't neutral eye candy sitting on top of good information. It actively costs the operator something: attention. Every second spent orbiting a 3D camera or waiting for a render to load is a second not spent noticing that a trend line has quietly been drifting for three days. Flashy isn't harmless. It can actively get in the way of clarity, not just fail to add to it.

To be clear: looking good is not the problem

I want to be careful here, because it's tempting to hear this as "good design is a waste of money" or "dashboards should be ugly and functional." That's not the point, and I don't believe it.

Attractive, well-designed interfaces are genuinely valuable. Good visual design reduces cognitive load, guides the eye to what matters, and makes a tool something people actually want to open instead of avoid. There is real craft in making dense information easy to scan. The problem isn't aesthetics; it's what we're using aesthetics to sell, and what we're using it to substitute for.

The failure mode is judging a dashboard's quality by how impressive it looks in a demo, rather than by how much faster and more confidently it lets a daily user make a correct decision. Those two things are not the same, and they're not even reliably correlated. Sometimes the most beautiful interface is also the most useful one. Sometimes it's the opposite. The point is that "impressive" and "valuable" are independent variables, and we've built a purchasing process that mostly measures the former.

The cost asymmetry nobody talks about

There's also a straightforward economic angle here that gets buried under the design conversation. A 3D rendered facility twin with flyover navigation is expensive to build and expensive to maintain. Every time equipment gets moved, replaced, or reconfigured, someone has to update the model. A well-designed table or trend panel with good historical context, thresholds, and comparison views is dramatically cheaper to build and keep accurate.

So the actual trade being made, often without anyone naming it explicitly, is: pay significantly more, for something that impresses the person who isn't going to use it daily, at the expense of the tool being less efficient for the person who is. That's not a design decision. That's a procurement incentive problem wearing a design costume.

AI is about to make this gap even more obvious

Here's the part I think is under-discussed right now. As more of these systems get an AI layer on top, an assistant that watches the sensor feed, flags anomalies, or answers questions about what's happening in the facility the value of the impressive layer doesn't just stay flat, it goes negative.

An AI reasoning over facility data doesn't want a 3D model to fly through. It wants clean structured data: readings, timestamps, thresholds, historical context, ideally in a plain table or a well-labeled time series. In fact, what's actually most valuable to it usually isn't a dashboard at all, it's direct API access to the underlying data, no interface required. Give a model a table of sensor values over time, or better yet a raw feed it can query, and it can spot the drift, the correlation, the anomaly. Give it a rendered 3D scene and, at best, it has to work a lot harder to extract the same signal from a format built for human spectacle rather than information density; at worst, the extra layer is just irrelevant to it.

So we're heading toward a strange split: the systems best positioned to actually consume and reason over operational data want exactly the plain, dense, tabular format that operators have quietly preferred all along, while the purchasing process keeps rewarding the format built to impress the people who never touch the data directly. If anything, AI is going to make the "flashy vs. functional" divide sharper, not smaller, because it removes any argument that the impressive version is also somehow the more "advanced" one. Plain, structured, information-dense is the advanced version now. It's what both the best human operators and the AI systems reasoning over the same data actually want.

---

Acknowledgments: We learn every day, and often from the least expected places. This article would not have been possible without the insights I gained while working alongside Ted Urbaniak and Venky Karuppanan among others.