Designing Interface Elements
Podcast · Designing Interface Elements

Designing Web Buttons and Forms That Work

by ZumiLabs · 48m
0:00 48m

Transcript

Made automatically from the audio, so it may contain mistakes. Click a line to play from there.

Host 1

You know, we have all been there at some point. You are walking up to a sleek, modern, I mean, just an incredibly expensive looking glass office door.

Host 2

Oh, yeah, I know exactly the kind you mean.

Host 1

Right. It has this perfectly symmetrical, beautifully brushed stainless steel vertical handle. It is just an architectural marvel.

Host 2

And it gives you absolutely zero clue whether you are supposed to push it, pull it, or slide it.

Host 1

Exactly. So you end up doing that awkward little dance, right? Where you pull, you smash your face against the glass, and then you quickly push while looking around to see if anyone noticed your sudden loss of basic motor skills.

Host 2

Yeah, the classic Norman door. It is it's the perfect physical embodiment of a design philosophy that prioritizes a minimalist aesthetic over fundamental human usability. I mean, when an object's physical appearance fails to communicate how it should be used, the designer has failed the user.

Host 1

It's so true. And welcome to today's deep dive into episode SO9E09, designing interface elements. This is part of our ongoing series on visual design for the web.

Host 2

Glad to be here for this one.

Host 1

Yeah, and just to give you, the listener, a quick recap of where we are with canopies, photography, aspect ratios, and twenty-four pixel icons beautifully unified back in SO9, episode eight. We now turn our attention to every single interactive control on the page so nobody ever hesitates over what to click or type into.

Host 2

Which is such a vital transition in our curriculum. Because that physical frustration you feel at the glass door, we are engineering the exact same frustration in the digital world right now.

Host 1

Oh, completely.

Host 2

Often under the guise of, you know, clean design.

Host 1

Right. Okay, let's unpack this.

Host 2

Okay.

Host 1

Because you really do see it everywhere. We are swimming in these ultra minimalist templates and AI generated user interfaces where flat decorative pill tags look exactly like clickable buttons.

Host 2

Yeah. They are indistinguishable.

Host 1

Or you are reading an editorial article and the actual navigation links just vanish inside paragraphs because someone somewhere decided underlines were too old school or visually noisy.

Host 2

Right. They just blend right in.

Host 1

And my absolute favorite, the one that makes me want to throw my laptop out the window. You start typing your email into a sleek borderless input field, and the second you type three letters, the placeholder label vanishes.

Host 2

Oh, the vanishing label.

Host 1

Right. Leaving you wondering if the form wanted your email, your username, or your phone number.

Host 2

Yeah, that vanishing placeholder is basically the digital equivalent of taking the handle completely off the door and just expecting the user to intrinsically know they need to press their palm against the glass.

Host 1

That is exactly what it is.

Host 2

We have systematically stripped away the visual signifiers. And without those signifiers, as Don Norman pointed out when he coined the term, users aren't navigating your interface. They are just guessing.

Host 1

Just guessing.

Host 2

And guessing requires an immense amount of cognitive load. It burns mental calories.

Host 1

Which means if we don't ground our component craft in fundamental human psychology, we are just making pretty pictures that fall apart the second a human being tries to use them. We are designing for dribble, not for reality.

Host 2

Exactly.

Host 1

So let's look at the literal mechanics of digital buttons. If a user is guessing, there must be a measurable cost to that confusion, right?

Host 2

Oh, there is. The data supports that assumption entirely. This isn't just theoretical UX philosophy. The Nielsen Norman Group, or NNG actually quantified this with a really extensive eye tracking study.

Host 1

I love when we get hard data on this stuff.

Host 2

Right. So they compared flat user interfaces where everything is stripped back minimal and lacks dimension against traditional interfaces with strong, unambiguous visual signifiers.

Host 1

Okay.

Host 2

And the results should be a wake-up call for any designer obsessed with extreme minimalism. In completely flat interfaces with weak signifiers, the task completion rate was over 20% slower.

Host 1

Twenty percent.

Host 2

Yes.

Host 1

I just want to pause on that number for a second. Because if you are running an e-commerce site or a SaaS platform, and you introduce a 20% delay in how fast someone can figure out how to add an item to their card or upgrade their plan, that is a massive loss in conversion. It translates directly to lost revenue.

Host 2

Absolutely. And if we connect this to the bigger picture, it is not just about the time lost. It is about the physical toll on the user. Watching the eye tracking heat maps from that study is incredibly revealing.

Host 1

What do they look like?

Host 2

Well, when an interface lacks clear signifiers, like a button that actually looks like a button with boundaries and contrast, the user's eyes physically dart wildly all over the screen.

Host 1

Like they're lost.

Host 2

Yeah. They are called saccades. The eye is hunting for clues. It scans text, evaluates shapes, and tries to deduce what might be interactive versus what is merely decorative.

Host 1

Wow.

Host 2

Whereas in a well-designed interface, the eye moves with calm precision, straight to the call to action, because its physical appearance screams, I am clickable.

Host 1

So the entire goal is to eliminate the hunt. We need to construct buttons that communicate with the user before, during, and after the interaction. Let's move away from just dropping a static rectangle in Figma.

Host 2

Right.

Host 1

A button is essentially a living organism in the browser, and it needs a life cycle.

Host 2

I like that framing. That life cycle consists of five core interactive states that must be explicitly designed and engineered.

Host 1

Let's walk through those.

Host 2

Okay, so the baseline state is enabled. This is your button at rest. It means the action is ready and available. It requires high contrast against the background and a highly legible label so it can be identified instantly.

Host 1

Okay, so that is the resting state, but a button needs to react when a user approaches it. If we think about desktop users, that reaction is the hover state, right? When the mouse pointer moves over the element.

Host 2

Exactly. Then we have the focus state.

Host 1

I assume this is the equivalent reaction, but for someone using a keyboard to navigate.

Host 2

You're hitting on a fundamental accessibility requirement there. Yes, focus is the visual reaction when a user has tabbed to the element using their keyboard. Or perhaps they are using a switch control device.

Host 1

Right. Okay.

Host 2

Finally, there's the pressed state when the system has actually registered the click or tap, and the disabled state, which we will definitely critique heavily later, when an action is temporarily unavailable.

Host 1

What really fascinates me is the neurological expectation tied to these reactions. The timing isn't arbitrary at all.

Host 2

No, it is entirely rooted in human perception. When a user tabs to a button or when they physically press it, that visual feedback must appear within 100 to 150 milliseconds.

Host 1

Which is essentially the threshold for human perception of instantaneous cause and effect. Oh, that makes sense.

Host 2

Yeah. They no longer perceive it as a direct reaction to their click. They think the system hitched or it missed their input entirely.

Host 1

I can visualize exactly what happens next. They think the button is broken, so they click it again.

Host 2

Exactly.

Host 1

And suddenly your database is dealing with duplicate forms of missions or a customer has accidentally ordered two expensive items, or the user is just deeply frustrated because your application feels sluggish and unreliable.

Host 2

And all of that because a CSS transition was set to 300 milliseconds instead of 100.

Host 1

That is wild. You know, a button without a proper hover or focus state reminds me of a broken elevator button.

Host 2

Oh yeah.

Host 1

You press it, the light doesn't turn on, and you instantly lose all trust in the machine. You know you pressed it, but the machine refuses to acknowledge you, so you just end up mashing it five more times with increasing aggression.

Host 2

It's so true.

Host 1

A digital button needs to light up. It needs to give you that micro feedback that says, I feel your input, I am processing your request. But let me play devil's advocate for a second.

Host 2

Go for it.

Host 1

We are living in a mobile first world. Everyone is tapping on flat glass screens. We don't have mice on our phones, so there is no hover state. Are we basically admitting that mobile UX is inherently inferior, or do we just need to compensate elsewhere?

Host 2

That is a really crucial friction point in modern design. Because mobile lacks a hover state, many designers assume they can just skip designing these intermediate states entirely, even for their desktop layouts. They think hover is dead.

Host 1

Right.

Host 2

But removing the hover state on mobile actually makes the pressed state and the focus state infinitely more critical.

Host 1

I see where you're going. Because on mobile, I don't get that reassuring pre click warning.

Host 2

Exactly. On a desktop, you move your mouse, the button darkens slightly, the cursor changes from an arrow to a pointer hand. You get the symphony of reassurance before you ever commit to clicking the mouse button.

Host 1

But on mobile.

Host 2

On mobile, you are flying blind. You tap a piece of glass and hope for the best. Therefore, the baseline enabled styling has to be absolutely unambiguous. It must look clickable from a mile away.

Host 1

Right.

Host 2

And the second they tap that 100 millisecond pressed state feedback is literally the only reassurance they get that they hit the right target before the next page loads.

Host 1

And I imagine this ties back to keyboard users on desktop too. The focus state basically becomes their hover state.

Host 2

It does. Focus states aren't just for desktop power users tabbing rapidly through data entry forms. They are essential for accessibility tools across all devices. If you neglect the focus state because you think it looks visually messy, you are effectively locking a massive segment of your users out of your application. They literally cannot see where they are on the page.

Host 1

Okay, so we have established what a button structurally is, why it needs to react, and the psychology behind those reactions. Let's take all of this theory and apply it to a real world mess.

Host 2

Let's do it.

Host 1

Let's look at the canopy app, which we started auditing in the previous modules. Right now, their top navigation bar is causing severe choice paralysis.

Host 2

It is incredibly noisy.

Host 1

Which, you know, is a common symptom of a product team where every department wants their feature front and center.

Host 2

Oh, I've been in those meetings.

Host 1

Right. So they have a logo, six different navigation links, a prominent brightly colored login button, an equally bright sign up button, and a book a demo button. All stuffed into a 60 pixel high top header. It's like walking into a crowded party where ten people are shouting your name at the exact same time. You don't know who to look at, so you just freeze.

Host 2

Yeah.

Host 1

This is Hicks Law in action, isn't it? The time it takes to make a decision increases with the number and complexity of choices. So we are going to calm this down using a framework that champions clarity over internal company politics we are paring that navigation down to just four clear destinations.

Host 2

Perfect.

Host 1

And let's talk about that login button because we are going to drastically reduce its visual footprint.

Host 2

We are stripping away its background color and its border entirely. It is going to become a quiet, simple text link.

Host 1

I assume that's because we need to separate the users by their intent. Existing users looking to log in already know what they want. They have high intent.

Host 2

Exactly.

Host 1

They don't need a massive glowing button begging for their attention.

Host 2

And and that gets to the heart of it.

Host 1

Exactly.

Host 2

And this brings us to a golden rule adopted by gov.uk.

Host 1

Oh, they're great.

Host 2

Yeah. They famously have some of the most rigorously tested usability standards in the world. So their rule is a single primary button per view. When you look at any given screen, there should be one obvious, undisputed next step that the interface wants you to take.

Host 1

Which means we have some serious work to do in Canopy's Hero section and their pricing tiers. Right now they have three solid, brightly colored buttons all stacked right next to each other. Start trial, book demo, view pricing. They are visually identical in their weight.

Host 2

Which forces the user to stop, read all three labels, weigh their options cognitively, and make a decision based purely on semantic text rather than being naturally guided by a visual hierarchy.

Host 1

That sounds exhausting.

Host 2

It is. To fix this, we implement a strict four-tier button hierarchy. This approach is heavily informed by both NNG principles and Apple's human interface guidelines.

Host 1

So we need to stop thinking about buttons as a monolith and start thinking about them as a descending scale of visual weight. Let's start at the top.

Host 2

The top of the scale is the filled primary button. This is reserved exclusively for the single main action you want the user to take on that view.

Host 1

Okay.

Host 2

It gets the highest visual emphasis, typically utilizing a solid background in your brand's core accent color. In Canopy's case, the business goal is likely driving signups, so start trial becomes the primary button. It must be the most visually heavy element in that cluster.

Host 1

Okay, so if start trial is the heavyweight, what happens to the runner-up actions? We don't want them to vanish, but they can't compete.

Host 2

Then you step down to tier two, the outlined or tonal secondary button. These are for actions of medium emphasis. They are important pathways, but they shouldn't fight the primary action for attention.

Host 1

How do we style that?

Host 2

This is achieved by giving the button a visible border, but a transparent background, or perhaps a very light, low contrast tonal background that recedes visually. For Canopy, book demo becomes a secondary button. The user still clearly recognizes it as an interactive target, but it allows the primary button to dominate the visual hierarchy.

Host 1

Got it. And then we drop down again for things that are just supplementary, tier three.

Host 2

Tier three is the text tertiary button. This has the lowest visual emphasis. It might look remarkably like a standard text link, but it behaves semantically like a button.

Host 1

Meaning?

Host 2

Meaning it performs an action on the current page rather than navigating you away to a new URL. View pricing, if it opens a modal or drops down a menu, could be moved to this tertiary style, ensuring it doesn't distract from the main conversion funnels.

Host 1

So we have primary, secondary, and tertiary. What is the fourth tier?

Host 2

Tier four is the destructive button. And Apple's Hig is very explicit about how this must be handled.

Host 1

I imagine this is for deleting things.

Host 2

Yes. A destructive button performs an action that can result in the permanent loss of data, like deleting an account, removing a record, or canceling a subscription. It should almost always use the system red color to communicate danger, and it should generally trigger a confirmation step. You never ever want a destructive button to share the visual styling of a primary or secondary action.

Host 1

So we take Canopy's three shouting buttons and we tame them. We turn them into one primary solid button for the trial, one secondary outline button for the demo, and we drop the pricing link to a quiet tertiary textile.

Host 2

It cleans it up beautifully.

Host 1

The visual hierarchy is now doing all the heavy lifting. The user's eye doesn't have to guess where to go. It is pulled naturally to the primary goal.

Host 2

What's fascinating here is the counterintuitive reality of conversion rate optimization.

Host 1

How so?

Host 2

Well, stakeholders often assume that if you make everything a giant bright button, more people will click. But by reducing the visual emphasis on those secondary and tertiary actions, you almost always increase conversions on the primary action. You are radically reducing the cognitive load required to make a choice.

Host 1

Less friction, less thinking, more action. I love that. But the visual styling is only half the battle. The friction can also come from the words inside the button itself. The UX copy. Right now, Canopy relies heavily on buttons that just say submit.

Host 2

Submit is a relic from the early days of HTML forms, and it is a terrible button label for anything other than perhaps a literal technical database entry.

Host 1

It's so vague.

Host 2

NNG warns heavily against generic labels because they destroy findability. They lack information sent and they fail to set expectations. When a user sees submit, they have to pause and scan the surrounding context to remind themselves what exactly they are submitting.

Host 1

It also feels incredibly demanding and machine-centric. You know, submit to the system. It doesn't sound like a benefit to the user at all.

Host 2

Instead, we need concrete verb object labels. What specific action is the user taking, and what is the object of that action? So we change submit to start 14-day trial. We change send to subscribe to newsletter. The label itself becomes a clear, explicit promise of what will happen the millisecond that button is activated.

Host 1

It is so much more reassuring. Now, while we were cleaning up the typography of interaction, we have to talk about text links. Specifically real editorial prose links that live inside a block of text, like in a blog post or a terms of service page.

Host 2

Right.

Host 1

The trend for the last few years has been to remove underlines entirely.

Host 2

Yes, there was a widespread movement to remove underlines to make paragraphs look cleaner, relying solely on a color change, like turning the link blue to indicate interactivity.

Host 1

But that's a problem, right?

Host 2

It's a massive accessibility failure. Relying on color alone doesn't work. Users with color vision deficiencies might not be able to distinguish a blue link from black body text. Furthermore, if a user is reading your site on a monitor with poor contrast or in bright sunlight, that color difference vanishes. The underline is the universal, unambiguous signifier for a hyperlink in body copy.

Host 1

Wait, I want to clarify something here. We don't underline the links in the main structural navigation at the top of the page, right?

Host 2

That is an important distinction. Structural links like a main navigation bar, a sidebar menu, or a footer layout derive their signifier from their placement, their grouping, and the surrounding layout.

Host 1

Because they are grouped together in a dedicated, recognizable navigation area, the user intrinsically understands they are interactive.

Host 2

Exactly. Underlining them there can indeed create unnecessary visual clutter. But the moment a link is embedded inside a paragraph of standard flowing text, it absolutely must have that underline.

Host 1

And we are just slapping a default HTML underline on it because that often slashes straight through the descenders of the letters, the tails of the lowercase Ps, G's, and Q's, which just ruins the type of graphic rhythm.

Host 2

It does.

Host 1

Since we already covered four-pixel and eight pixel spatial systems in the previous episodes, we know how to handle micro adjustments. We restort a crisp underline, but we offset it. We drop that underline by two to three pixels below the baseline. It keeps the typography beautifully legible while preserving the signifier.

Host 2

It is a small detail, but it elevates the entire reading experience. It shows a deep respect for both aesthetics and usability.

Host 1

Okay, we have optimized discrete actions.

Host 2

We know how a single button should act, what it should say, and how an inline link should look. But let's transition to a much more complex element that causes headaches for both designers and developers.

Host 1

Oh, I know where this is going. What happens when a whole block of content needs to act as a gateway? I'm talking about cards.

Host 2

Cards are ubiquitous in modern web design. They are fantastic for grouping related content. An image, a title, a short description snippet, maybe a date. Yeah. But they present a massive technical and usability dilemma when you need that entire card area to be clickable.

Host 1

Right. We call this the Matrioshka nesting doll dilemma. This is a term popularized by the accessibility research of Hayden Pickering and standards from the US web design system. Let's walk through the friction here. You have a card that represents a blog post. Naturally, for mobile users especially, you want the user to be able to tap anywhere on that large card to navigate to the post.

Host 2

Sure.

Host 1

The lazy traditional way to build this is to take the HTML anchor tag, the tag that creates a link, and just wrap it around the entire card structure.

Host 2

Which seems perfectly logical on the surface. You want the whole box clickable, so you wrap the whole box in a link. But structurally, under the hood, it is a disaster.

Host 1

Why is it a disaster?

Host 2

When you wrap a large container full of diverse block-level content inside a single inline link tag, you create severe accessibility and usability issues.

Host 1

Let's put ourselves in the shoes of someone using a screen reader for a second. Imagine encountering that wrapped card. Because the entire card is coded as one giant link, the screen reader doesn't just read the title, right?

Host 2

No, it reads everything inside that wrapper as one massive verbose link label. It is exhausting for the user.

Host 1

Oh no.

Host 2

A screen reader user navigating a page often wants to tab rapidly through links to find the content they need. If they land on a nested card, they might hear Link Designing Interface Elements by Jane Doe, October 14th, image of a computer, 10 common pitfalls to avoid when designing components.

Host 1

By the time it finishes reading the description snippet, the user has completely forgotten what the title was. They just wanted to hear link, designing interface elements. Not an audiobook of the entire card's contents.

Host 2

Exactly.

Host 1

And it's not just a screen reader issue, is it?

Host 2

Not at all. Wrapping the whole card in a link completely breaks text selection for mouse users. If a user tries to click and drag to highlight the description text, perhaps they want to copy a quote, they end up accidentally dragging the entire link component instead. It feels broken.

Host 1

And heaven forbid you try to put a second link inside that card. Like, say you have a little category tag in the corner that says UI design, or you want the author's name to link to their bio.

Host 2

You cannot put a link inside a link. The HTML specification forbids it. The browser's rendering engine won't know how to handle the nested interaction, and assistive technologies will completely fail to parse it. It is a structural dead end.

Host 1

So we have this massive tension. We want the ergonomic luxury for mouse and touch users to be able to click anywhere on the big generous target area of the card. But we desperately need a clean, semantic structure for screen readers, and we need text selection to remain intact.

Host 2

It's a tricky balance.

Host 1

This is where we deploy what I like to call the audio first solution. Without getting bogged down and reading raw CSS syntax aloud, I like to explain this technique using the analogy of the invisible tablecloth.

Host 2

The technique is formally known in the industry as the stretched link pattern. And your tablecloth analogy captures the mechanism perfectly.

Host 1

Okay, picture the card sitting on a table. You have your image, your heading, your description text. Instead of wrapping the whole thing in a link, you only make the main heading, the actual title of the blog post, the HTML link. Everything else is just normal text.

Host 2

And structurally, this is exactly what the screen reader wants. When it tabs to the card, it lands purely on the heading. It announces link. Designing interface elements. Clean, concise, perfect information sent.

Host 1

But if we leave it like that, a mouse user clicking the bottom corner of the card gets nothing. They just click empty space.

Host 2

Right.

Host 1

So here's where the magic happens. Using a neat little CSS trick with pseudo-elements, which you can basically think of as a ghost element the browser creates just for styling purposes. We take that link that lives on the heading and we stretch an invisible, clickable area out of it. We pull it down and stretch it over the entire boundary of the card. We literally drop a clear, invisible tablecloth over the whole thing.

Host 2

The mechanics are brilliant because it satisfies every distinct user need.

Host 1

Tell me more.

Host 2

The mouse user clicks the invisible tablecloth anywhere on the card, and because that tablecloth is technically tied to the heading, it triggers the navigation. They get their massive satisfying touch target. Meanwhile, the screen reader only sees the semantic link on the heading, getting a single clean tab stop.

Host 1

And what about the text selection issue? If the tablecloth is covering the text, how can I highlight it?

Host 2

This comes down to how you manage your Z indexes, meaning the CSS rules that dictate how elements stack on top of each other, like layers of paper. You can configure the layering so that the text of the description sits technically above the invisible stretched link while the rest of the card sits below it.

Host 1

Oh, that's clever.

Host 2

It is. The user gets the ergonomic benefits of a massive click target without compromising the integrity of the document structure or text selection.

Host 1

It is one of those solutions that just feels like a magic trick the first time you implement it. It completely solves the Metryoshka dilemma. You don't have links inside links, you just have one clean heading link stretched to cover the container.

Host 2

This level of component craft is what separates superficial design from exceptional inclusive engineering. We aren't just making it look good, we are engineering it to be resilient for every type of human interaction.

Host 1

Okay, let's keep this momentum going. We have optimized clicking. We have made sure our buttons and cards are bulletproof. Now we have to optimize typing.

Host 2

Forms.

Host 1

Yes, and this brings us to one of my biggest pet peeves on the modern web. We are looking at Canopy's sign-up form. Currently, it is a two-column layout, and it relies entirely on placeholder text instead of persistent labels.

Host 2

The placeholder only form. It is the absolute darling of minimalist designers everywhere who want things to look uncluttered, and it is an unmitigated usability nightmare.

Host 1

It really is.

Host 2

The Nielsen Norman group has actually published an entire research report detailing seven distinct reasons why placeholder text used as labels is actively harmful to users.

Host 1

It's like writing a form in vanishing ink. Well, let's walk through the friction here. The placeholder says, enter your email address in light gray text inside the input box. The second you click into that box and type the letter J for your name, poof. The label is gone.

Host 2

And that leads directly to NNG's first and most critical point. Placeholders severely strain short-term working memory. We have to remember Miller's Law, which states that the average person can only hold about seven items in their working memory at once. Right. Furthermore, filling out a form is rarely an activity that gets 100% of a user's undivided attention. You start typing your email, your phone buzzes with a text message, you look away to read it, you look back at the screen, and you have an input box with JAS in it. The label is gone. Was this the email field or the username field or the first name field?

Host 1

You literally have to highlight where you just typed, delete it, click outside the box to make the vanishing ink reappear, read the label again, and start over. It is infuriating.

Host 2

Which brings us to the second point. It prevents users from checking their work. Before you hit that final submit button, you should be able to scan down the form and verify that the data you entered matches the requested fields. Without persistent labels, you are just looking at a list of answers without the questions.

Host 1

Right. And what if the browser's autocomplete jumps in? Autocomplete is notorious for putting the wrong info in the wrong box. It eagerly drops your phone number into the zip code field. If there are no persistent labels to check against, you might just blindly submit a completely botched form.

Host 2

Exactly. NNG also notes that placeholders make error recovery nearly impossible. If the system throws a validation error saying, invalid format, but the label has vanished, the user doesn't know what format was originally expected.

Host 1

And let's talk about the visual design of placeholders themselves. They are almost universally styled in that faint wispy gray.

Host 2

This is a major accessibility violation. To meet standard WCAG guidelines, text generally needs a contrast ratio of at least 4.5 to 1 against its background. Default placeholder text in most browsers almost never meets this requirement.

Host 1

No, it's always so light.

Host 2

It is incredibly hard to read for anyone with visual impairments or even just someone using their phone out in bright sunlight.

Host 1

But here is the craziest part that I've seen in user testing. Sometimes people don't even realize the field is empty.

Host 2

Yes, this is a very common and frustrating observation. Because there is text physically sitting inside the input box, users scan the form and their peripheral vision registers that field as already filled out. They mistake the placeholder for a pre-filled default value.

Host 1

So they just skip the field entirely, hit submit, and immediately get hit with an error.

Host 2

Or conversely, they click into the field and actually try to manually delete the placeholder text before they start typing because they think it's real text they need to backspace over. The interaction cost is just skyrocketing.

Host 1

It is a perfect storm of bad usability, all sacrificed at the altar of making the form look slightly less cluttered when it's completely empty.

Host 2

Absolutely.

Host 1

So how do we rebuild Canopy's signup form to fix this? We are killing the vanishing ink. What is the gold standard layout here?

Host 2

Following rigorous benchmarks from both the W3C and the Baymart Institute, who conduct massive usability studies on e-commerce checkout flows, we are going to rebuild this form into a single column flow.

Host 1

I assume that's not just to save horizontal space. Why is a single column better than two columns side by side?

Host 2

Two column forms force the user's eyes to zigzag across the page. Left, right, down, left, right, down. It disrupts the natural linear reading flow and drastically increases the cognitive load required to track which field comes next.

Host 1

I can see how that gets confusing.

Host 2

It significantly increases the chance that a user will completely skip a field. A single column ensures a smooth, predictable, linear progression straight down the page.

Host 1

Okay. Single column flow. And the labels, since we aren't putting them inside the box anymore, where do they go?

Host 2

We use persistent, highly visible labels placed directly above the input field.

Host 1

Above the input, not to the left of it aligned horizontally.

Host 2

Here's where it gets really interesting. Placing labels above inputs is the most resilient and scalable layout for modern web design because it directly caters to mobile viewports.

Host 1

Well, that makes sense.

Host 2

If you put a label to the left of an input, you run into severe horizontal space constraints the second that form is viewed on a narrow phone screen. Either the label text gets squished and awkwardly wraps to three lines, or the input box becomes too small to actually see what you were typing.

Host 1

Right.

Host 2

By stacking the label on top of the input, the layout works perfectly on a massive desktop monitor and scales down flawlessly to an iPhone screen without ever causing horizontal scrolling or layout shifts.

Host 1

That makes so much sense. And we know from our prerequisite design systems in previous episodes, we are going to use very specific spacing here.

Host 2

Yes.

Host 1

We're going to put that top label about four to six pixels above the input box. We want it physically tight to the input, so the psychological law of proximity tells the user without a doubt, this label belongs exclusively to this box. If you float the label too far away, it gets confusing which box it belongs to.

Host 2

Precisely. And the input box itself needs to be substantial. We are aiming for a 44 pixel tall input box to ensure it is a comfortable, generous touch target for a thumb on a mobile device.

Host 1

And we are giving that input box a solid, clear border. We need a minimum three to one contrast ratio for that boundary so the input actually, it looks like a physical input well. It shouldn't look like a faint suggestion of a rectangle. It needs to be a defined interactive space.

Host 2

This brings us to a critical transition in component design. Which is we have spent all this time perfectly crafting the visual appearance of the form. But even if we have the perfect layout, single column, top labels, high contrast borders, the actual behavior of these form controls can still completely trap users if we rely on generic out-of-the-box template logic. We need to talk about validation and edge cases.

Host 1

Yes. This is where we transition from how the interface looks to how it logicates the user. And I want to bring up a trend that has been driving me absolutely insane lately.

Host 2

Let's hear it.

Host 1

You spend five minutes filling out a long form, you get to the very bottom, and the submit button is just grayed out. Disabled. And the form offers absolutely no explanation as to why.

Host 2

The trap of the disabled submit button. It is a usability dead end, and yet it is incredibly common in modern web development frameworks like React or Vue that validate the form state in real time.

Host 1

Why did they do it?

Host 2

The underlying logic from the developer's perspective is the form state is currently incomplete or invalid, therefore I will protect the database and not allow the user to submit it.

Host 1

Which sounds perfectly logical to a computer, but to a human being trying to accomplish a task, it's a brick wall. You are forced to scroll back up the form, squinting, trying to figure out which field you missed. Did I format my phone number wrong? Was the password missing a special character? Did I miss a tiny required checkbox?

Host 2

Exactly.

Host 1

A disabled button cannot communicate what is wrong. It only communicates that you have failed. So what's the alternative?

Host 2

The far superior user-obsessed approach is to leave the submit button enabled at all times. If the user clicks it while the form is incomplete, you intercept that click via JavaScript, prevent the actual submission to the server, and then explicitly highlight the fields that need attention with clear inline error messages.

Host 1

I love that.

Host 2

You use the user's action to provide instructional, helpful feedback rather than just locking the door and throwing away the key.

Host 1

It's the difference between a teacher silently crossing their arms and saying, You got something wrong, but I'm not going to tell you what it is until you guess it. Versus a teacher saying, Here, let's look at question number four together. It looks like you missed a decimal.

Host 2

Perfect analogy.

Host 1

But speaking of error messages, there's a distinctly wrong way to deliver those two. I call it premature form yelling.

Host 2

I love that phrase. Explain what you mean by premature yelling.

Host 1

Okay, picture this. You click into the email field, you type the letter J, instantly, the entire form flashes bright red, a warning icon appears, and it screams at you, invalid email format, missing at symbol.

Host 2

Oh, I've seen that.

Host 1

And I'm just sitting there thinking, I know it's missing the at symbol. I literally just started typing. Give me a second to finish my name.

Host 2

It is incredibly aggressive, and it causes a literal measurable cortisol spike in users. It induces anxiety. This happens when forms are lazily configured to validate on input, meaning they run their validation script after every single keystroke. Right. If you are halfway through typing an email address, of course it is technically invalid.

Host 1

I assume the fix here is just a matter of timing. We need the system to be a little more patient.

Host 2

Validation should be calm and it should wait until you have finished your thought. We achieve this by triggering the validation on blur.

Host 1

On blur meaning when the user looks away.

Host 2

Meaning the exact moment the input field loses focus, the user types their email in peace. The system remains quiet. When they are done, they hit the tab key, or they physically click into the next field.

Host 1

Okay.

Host 2

That action of leaving the field is known as the blur event. That is the explicit signal that the user has finished entering their data. That is the exact moment the system should check if the email is valid. If it isn't, then you display the calm, helpful recovery message below the input.

Host 1

It's just common courtesy. Let me finish my sentence before you correct my grammar. Now, regarding the inputs themselves, there's a brilliant guideline from gov.uk about the actual physical width of the input fields. We shouldn't just make every single input 100% of the column width to make the layout look perfectly flush, right?

Host 2

No, we absolutely shouldn't. The physical width of the input should visually match the expected length of the answer.

Host 1

I assume that's not just to save screen real estate. Does the physical size of the box act as a psychological hint for what you're supposed to type?

Host 2

You're hitting on a fundamental psychological friction point. It is a subtle but incredibly powerful visual cue that sets expectations.

Host 1

Give me an example.

Host 2

If you're asking for a three-digit security code and you provide an input box that stretches entirely across a 27-inch desktop monitor, it creates cognitive dissonance. The user hesitates, they wonder if they are supposed to type more, or if they misunderstood the question. Right. By deliberately sizing a postal code field to be short and an email field to be wide, you are subconsciously confirming to the user that they are in the right place, entering the right type of data.

Host 1

It is entirely about reducing that micro hesitation. And speaking of types of data, we need to issue a massive warning about asking for credit card numbers or zip codes. I see developers all the time using the HTML number input for this. You know, when the code literally says input type equals number.

Host 2

Yes. This is a very technical trap that catches a lot of developers who are trying to be semantically correct, but end up ruining the UX. Because a credit card is made entirely of numbers, they assume they should use the number input type.

Host 1

It seems logical at first glance.

Host 2

But the number input type in HTML was designed specifically for quantities. How many tickets do you want to buy? What is your age?

Host 1

Right. A credit card isn't a quantity. I don't have 45 trillion credit cards. It is an identification string that just happens to consist solely of digits.

Host 2

Correct. And the primary danger is that the number input often comes with built-in UI mechanics from the browser, like little up and down stepper arrows inside the box.

Host 1

Oh, that's annoying.

Host 2

More dangerously, if a user clicks into a number field, types their number, and then accidentally uses the scroll wheel on their mouse or swipes two fingers on their trackpad, the browser will often increment or decrement the number.

Host 1

Oh well. So I type in my 16-digit credit card number perfectly. I leave my cursor in the box while I scroll down the page to find the submit button, and my credit card number silently changes by one digit in the background.

Host 2

Exactly. The final digit goes from a four to a five, the user hits submit, the payment fails, and the user has absolutely no idea why, because they know they typed it correctly.

Host 1

That is awful.

Host 2

For things like credit cards, postal codes, or account numbers, you should always use the standard text input. But to ensure mobile users get the correct keyboard, you configure the input mode attribute to numeric. This pulls up the large numeric keypad on their phone, giving you the ergonomic benefit without the dangerous scrolling trap of the number input.

Host 1

Okay, we are getting deep into the tool workflows here, which is our next major section. To make these designs actually work in the real world, we have to respect how the browser fundamentally interprets the controls we are building. And this brings up the great semantic divide buttons versus links.

Host 2

It is arguably the most fundamental structural distinction in web development. And yet it is constantly violated by well-meaning designers and developers. According to the MDN Web Docs, which serve as the absolute source of truth for web standards, a button and a link serve entirely different foundational purposes, regardless of how they are styled to look visually.

Host 1

I am going to push back here on behalf of every designer working in Figma right now. If I design a beautiful blue rectangle with perfectly rounded corners and bold white text, and it looks exactly the same visually whether the developer codes it as a button or a link. Okay, break it down for me.

Host 2

Let's define the rigid rule. A button tag is used to perform an action on the current page. It submits a form, it opens a popover menu, it closes a dialog box, it plays a video. An anchor link tag is used exclusively to navigate you somewhere else. It changes the URL. It takes you to a new page or a different section of the current page.

Host 1

Okay, but if I use a button tag to link to my about us page, it still takes them there when they click it. Why is that bad?

Host 2

Because you break fundamental web expectations and strip away user control. If you use a button tag for navigation, a user cannot right click it and select open a new tab.

Host 1

Oh, I do that all the time. I middle click things constantly on news sites to open them in background tabs while I keep reading the main page.

Host 2

Millions of power users do. But a button tag doesn't have an underlying URL in the way a link does. It just fires a JavaScript event. So if you build a navigation menu using buttons instead of links, you completely break the user's ability to browse the web the way they prefer.

Host 1

That makes total sense.

Host 2

Conversely, if you use a link tag to submit a form or open a modal, you break accessibility expectations. A screen reader will announce the element as a link, setting the mental expectation that the user is about to leave the page. When they activate it and a pop-up appears instead, they are entirely disoriented.

Host 1

It is a betrayal of trust. The element looks like it will do one thing, but structurally it promises another, and functionally it does a third. We have to honor Jacob's law here.

Host 2

Jacob's Law of Internet user experience.

Host 1

It is a cornerstone principle. It states that users spend most of their time on other sites. This means that users prefer your site to work the same way as all the other sites they already know. They bring their expectations from Amazon, Google, and Apple directly to your product. So whether your brand aesthetic is a warm editorial style with elegant serifs and soft earth tones, or a crisp Swiss technical utility style with sharp corners and rigid grid lines, or a tactile craft brand style with heavy shadows and gradients. None of that visual flair excuses you from how the interaction is expected to behave under the hood.

Host 2

Precisely. You implement these diverse visual styles in Figma through component variants, you map them to web flow or framer states, and you write the CSS to make them look beautiful. But the underlying mechanics, the hover states, the focus rings, the semantic HTML tags must remain rigorously consistent with the broader web.

Host 1

I see.

Host 2

If we connect this to the bigger picture, standardizing these interactive components is actually incredibly liberating for a design team.

Host 1

Liberating. Most designers feel restricted by standards. How is it liberating?

Host 2

Because it frees up your design budget, both your literal financial budget and your team's cognitive bandwidth. You don't need to reinvent the wheel on how a text input functions. You don't need to spend hours designing a bespoke, never before seen interaction for a drop-down menu that ultimately just confuses people. Right. You use the standard bulletproof mechanics and you spend your creative energy on the imagery, the layout, the storytelling, and the brand messaging.

Host 1

You innovate on the content, not on the plumbing. If you buy a house, you want the light switches to work exactly like every other light switch in the world, you don't want a puzzle.

Host 2

Exactly.

Host 1

Okay, this has been an incredibly deep dive. We've covered psychology, visual design, accessibility, and HTML semantics. To make sure you, the listener, can actually use all of this tomorrow. We are going to round up all these exact specs into a mental cheat sheet. Let's do a rapid-fire spec recap of our reference thresholds.

Host 2

Sounds good. I'll start with hierarchy. You must strictly enforce one primary call to action button per view. Tame all other actions into secondary or tertiary styles so they don't compete.

Host 1

Next is sizing. Your touch targets for buttons need to be at least 44 by 44 pixels, as per Apple's HIG. Do not settle for the absolute bare minimum 24 by 24 pixel floor that WCAG 2.28 technically allows. 24 pixels is just too small for a human thumb in a hurry. Go 44.

Host 2

Moving to form inputs. The input fields themselves should be 40 to 48 pixels in height. Ensure the boundary border has at least a three to one contrast ratio against background, so the input well is clearly visible as an interactive area.

Host 1

For proximity and layout, position your persistent form labels four to six pixels directly above the input field, and make sure the gap between different form fields is much larger, like 16 to 24 pixels. The label must clearly visually belong to its specific input.

Host 2

Regarding editorial links. Embedded prose links need an underline, and that underline must be offset by two to three pixels so it clears the descenders of the typography, keeping the text readable.

Host 1

And finally, interactive states. Every single button and interactive element must have six distinct states planned out in your design system. Enabled, disabled, hover, focus, pressed, and a loading state when appropriate to reassure the user that the system's working.

Host 2

Having this spec sheet isn't just about following rules for the sake of compliance. It is about building a system of trust. When your interface is predictable, legible, and responsive, the user stops thinking about the interface entirely and focuses solely on the task they are trying to achieve.

Host 1

So what does this all mean? It means you now have a foolproof system to guarantee user confidence. No more norm and doors in your UI. No more vanishing placeholders or broken middle clicks. And because we believe in putting theory into immediate practice, I have a challenge for you, the listener.

Host 2

Oh, I think they are ready for it.

Host 1

I challenge you to a 15-minute cleanup on your current project. Go to the sign-up card or the primary form you are working on right now. Apply these exact thresholds, strip out the placeholders, move the labels to the top, check your three to one contrast on the borders, verify your focus states are visible for keyboard users, and brutally enforce that single primary button rule.

Host 2

I love this idea.

Host 1

I guarantee you, in 15 minutes, that form will look 100 times more professional and perform drastically better in user testing.

Host 2

It is remarkable how quickly an interface sharpens and feels premium when you methodically remove ambiguity.

Host 1

It really is. And this brings us perfectly to the horizon of our series. Now that we have mastered the structure of these static flat controls, it is time to literally elevate them.

Host 2

Can't wait for the next one.

Host 1

Next up in episode S09E10, Depth Surfaces and Motion, we are going to take these perfectly structured buttons and cards and give them subtle tactile physics. We are going to explore how drop shadows, Z-index layering, and micro interactions can make a flat screen feel like a physical tangible object that responds to touch.

Host 2

But before we leave today's topic of interactive states, I want to leave you with a final slightly provocative thought to ponder. Go a piece of physical glass.

Host 1

Right, the whole hundred to hundred and fifty millisecond response time based on physical action?

Host 2

Yes. But as we move rapidly towards spatial computing, look at Vision OS, where Apple has already noted that hit targets must jump to a massive 60 by 60 points simply because they rely on eye tracking and subtle pinch gestures. Or even looking further out to future neural interfaces, how will the cognitive load of a button state evolve when the physical act of clicking disappears entirely?

Host 1

Oh wow. If the computer just intrinsically knows I'm looking at a button and thinking about activating it.

Host 2

Exactly. When you don't even tap your finger, will our digital environments need to physically feel different to our brains? Perhaps through advanced haptics or direct neural feedback just to confirm an action has been registered. If the Norman door loses its physical handle entirely, and the glass itself just reads your intent, how will the system let you know that you've successfully opened it?

Host 1

That is a wild thought to end on. We aren't just designing visual hover states anymore. We are moving toward designing the hover state for the human brain itself.

Host 2

It's the next frontier.

Host 1

We have a lot of work to do. Thank you so much for joining us on this deep dive into the foundations of design. Keep experimenting, keep auditing those forms, take the 15 minute challenge, and as always, keep diving deep.

Save to playlist

Manage playlists →

Share