Your UI Is Exhausting

TL;DR: Cognitive load kills patience faster than teams think. Every extra label, choice, and pointless step burns mental fuel the user did not plan to spend. People do not quit because they are lazy. They quit because your UI is exhausting.

A user opens your product with a task in mind. They hit one small ambiguity, then another. A label that needs a second read. A button whose purpose isn’t obvious until they click it wrong. A flow with one extra step that nobody can explain but removing it felt risky, so it stayed. Nothing breaks. Everything costs. The mental effort adds up, and at some point the cost exceeds what the task is worth to them. They close the tab. No report, no feedback, no explanation. They just stop.

You’ll read it in your analytics as abandonment. What it actually is, is exhaustion.

Four slots. That’s it.

For a long time, designers quoted a single number: seven. George Miller wrote a paper in 1956 suggesting people could hold roughly seven items in short-term memory, give or take two. The design world turned it into law. Menus capped at seven. Forms structured around seven fields. It showed up in UI guidelines for decades. What didn’t travel with the number was Miller’s own caveat. He called it rhetorical, a rough approximation, not a measurement. Designers took a ballpark and calcified it into a rule.

Nelson Cowan went back through the evidence in 2001 and found the real ceiling is closer to four. Four chunks, held and worked with simultaneously, before things start falling out. Not seven. Four. Which means the interface you built assuming users could track seven things in parallel was already over budget before anyone opened it.

John Sweller described three types of cognitive load. Intrinsic load is the difficulty inside the task itself. Filing taxes is hard, and design can’t change that. Germane load is the useful thinking: a user working something out, building understanding, moving forward. Both of those are load you want. They’re the point. The third type is what designers accidentally create.

The extra mental effort the interface creates is the effort caused by how information gets presented. Unclear hierarchy. Redundant steps. Labels that made sense in a product spec but mean nothing to a real person. This load occupies working memory the same way everything else does. As Nielsen Norman Group put it: “When the amount of information coming in exceeds our ability to handle it, our performance suffers.” The brain doesn’t distinguish between effort that serves the user’s goal and effort that doesn’t. It just runs a tab. When the tab gets too high, the task stops. The user isn’t weighing whether to continue. They’ve already left.

What TurboTax actually did

I want to be careful with this example because people draw the wrong lesson from it.

TurboTax dominated US consumer tax filing for years, and the design is genuinely worth studying. The task it helps people through is one of the hardest civilian things there is. IRS forms are dense, conditional, and written as if the reader already knows what the instructions mean. TurboTax didn’t simplify the tax code. It simplified the experience of moving through it: one question per screen, plain English where jargon could be cut, a progress indicator, defaults that removed decisions users didn’t need to make from scratch. The insight wasn’t sophisticated. It was just honest about what the task already cost and choosing not to add to it.

Competing products that stacked questions on a single screen, left technical terms unexplained, or made users navigate between sections without a clear thread saw more people drop off. Not because the task was harder. Because the extra mental effort was higher. Users weren’t comparing products. They were running out of capacity before they reached the end.

Worth being clear: Intuit’s record outside the product is not good. Their lobbying against free federal filing and their behavior around the free tier are well documented. A design principle doesn’t need a virtuous company behind it to be true. But you shouldn’t use TurboTax as proof that simplicity always wins or that users are fragile. The lesson is narrower than that: the task already costs something, so the interface shouldn’t cost extra.

The audit nobody actually runs

Pull up your most-used screen and look at it like you’ve never seen it before. Count what you’re asking a user to hold in mind before they can do the thing they came to do. Count the labels that need interpretation, the steps that exist because removing them seemed risky rather than because they serve the user, the options nobody touches but nobody cut. Each one is a withdrawal from a four-slot account. Most of them are residue: added when adding felt like progress, kept because nobody went back to ask what they were costing the person on the other side.

One question decides it: does this require effort from the user that serves the user’s goal? If the answer is no, it goes. Not as an aesthetic call. Not because sparse is better than full. Because the budget is four slots and your interface doesn’t get to spend most of it before the task even begins. The next hour you spend on a screen is better spent removing something than adding it.

Your users aren’t abandoning because they aren’t trying. You spent their budget before they got to the part that mattered.