A handful of web design habits spent last year pretending to be innovation. They were effects, not decisions, and they cost sites the two things that actually earn a visitor's trust: speed and clarity. Here's what we're leaving behind, what the research says about why, and what we're building toward instead.
The year the effects outran the experience
2022 was a good year for web design if you liked watching pages think. Hero sections opened with looping video before a single word of copy loaded. Scroll position triggered choreography instead of content. Cookie banners buried "reject all" three menus deep while "accept" sat one thumb-tap away. Somewhere along the line, a lot of teams started building for the demo reel instead of the person actually trying to get something done.
We get the appeal. Effects are fun to build and fun to show off in a pitch deck. But a website has one job before it has any others: let the person who showed up find what they came for, fast, without a fight. Every trend on this list failed that job in the same way. It asked the visitor to wait, watch, or work harder than they should have had to, in exchange for a feeling the site’s owner wanted them to have. That’s a backwards trade, and going into 2023, we’re not making it anymore.
This isn’t a call to strip every site down to Times New Roman on white. Craft and personality still matter, and always will. The line we’re drawing is between design that serves the visitor and design that performs for the sake of performing. Here’s what crossed it.
Autoplay hero video: the tax nobody asked to pay
Autoplay video heroes were everywhere in 2022. They look expensive. They also load slow, and speed is not a nice-to-have anymore, it’s the front door.
Google’s own Core Web Vitals research, built on real user data from Renault’s global site, found that a one-second improvement in Largest Contentful Paint led to a fourteen percentage point drop in bounce rate and a thirteen percent increase in conversions (Google, "Renault"). A hero video is very often the single heaviest asset on a page, and heavy assets are exactly what drag Largest Contentful Paint into the range where people leave before they arrive. That’s not a design opinion. That’s what happens when a ten-megabyte file sits between a visitor and your headline.
“1 second LCP improvement can lead to a 14 percentage points (ppt) decrease in bounce rate and 13% increase in conversions.”
Google, "How Renault Improved Its Bounce and Conversion Rates by Measuring and Optimizing Largest Contentful Paint," web.dev
There’s an accessibility cost too, one that predates 2022 but that a lot of sites still ignore. The Web Content Accessibility Guidelines have required, since version 2.0, that any moving content lasting more than five seconds come with a way to pause it (Success Criterion 2.2.2, "Pause, Stop, Hide"). An autoplay hero that loops without controls isn’t just heavy, it’s out of spec. Motion for motion’s sake was never free. In 2022 we finally got the receipts.
Going into 2023, we’re building hero sections that load a sharp static image or a compressed, muted, user-controlled clip, and we’re putting the words first. If a video earns its place, it needs to prove it, not just exist because everyone else’s did.
Parallax scrolljacking: an effect users never asked to feel
Parallax had a long run, and it’s earned its retirement. Nielsen Norman Group’s usability research on parallax scrolling found something worth sitting with: users in testing sessions didn’t stop to admire the layered scroll effects, didn’t scroll back up to watch them again, and in follow-up interviews reacted to parallax pages the same way they had five years earlier, with plain indifference (Sherwin). The effect a designer spent hours polishing simply didn’t register as a reason to care.
“They aren’t ‘wowed.’ User”
testing participants don’t stop to comment on how cool the effects are; they don’t scroll up and down for fun, to watch the effects bring the page to life." — Katie Sherwin, "What Parallax Lacks: A Usability Study," Nielsen Norman Group
Worse, the same research flagged real physical discomfort. Some participants reported disorientation when scroll speed and animation speed fell out of sync, the same effect that pushed Apple to ship a "Reduce Motion" setting in iOS 7 after users reported vertigo and nausea from parallax-heavy interfaces (Sherwin). That’s not a trend overstaying its welcome. That’s a trend that was never fully welcome to begin with, dressed up as sophistication.
Scrolljacking, where a site hijacks native scroll behavior to force a scripted sequence, carries the same problem with an added insult: it breaks the one interaction every visitor already knows how to do. We’re not asking people to relearn scrolling to see our work. The backbone of good web design is respecting what the visitor already knows.
Dark-pattern cookie banners: consent theater
The last habit we’re done with isn’t a visual trend, it’s a manipulation dressed as one. Cookie banners that make "accept all" a single bright button and "manage preferences" a maze of toggles and sub-menus aren’t design decisions. They’re a bet that visitors will give up before they find the exit. Some of the most common patterns, pre-checked opt-in boxes, color contrast that hides the reject option, false urgency copy, do the same job dark patterns have always done: they trade a visitor’s trust for a marginal data win.
We build consent flows that treat "no" and "yes" as equally easy to find, equally sized, equally worded. It’s a small thing on a page and a defining thing about who built it. Overlooked details like this are exactly where conviction shows up or doesn’t.
What actually matters going into 2023
Strip away the trend chasing and three things are left standing: speed, accessibility, and clarity. They were never glamorous, and that’s exactly why teams kept skipping past them for something flashier. Going into 2023, they’re the whole brief.
Speed means treating Core Web Vitals as a design constraint from the first wireframe, not a performance audit bolted on after launch. Accessibility means building to WCAG as a floor, not a checklist item for legal to worry about later. Clarity means every scroll, every animation, every autoplay decision has to earn its place by helping the visitor, not by impressing whoever’s in the room during the pitch.
None of that is a lower bar than the effects we’re retiring. It’s a harder one. Building something fast, accessible, and clear takes more discipline than layering on a parallax plugin, not less. That’s the kind of hard we want. Safe doesn’t survive here, and neither does slow.
