Why the best UX in 2026 knows when to slow users down — and how AI is forcing designers to make that call more deliberately.
The head of airport operations at Singapore Airlines once told a designer: make the experience forgettable. Not a joke — a doctrine. Thousands of passengers move through that airport daily, and joy lives in the flight itself, not the boarding process. Friction, in that context, is a failure state.
But here’s the problem: the design industry is generalising that lesson somewhere it doesn’t belong.
When Frictionless Becomes Forgettable in the Wrong Way
As Darren Yeo argues in UX Design, the obsession with removing all friction has created a paradox: when every interaction disappears, nothing registers. Users complete flows, convert, churn — and remember nothing about why they chose you. That’s fine for airport queues. It’s a slow disaster for brand loyalty.
The data layer makes this especially legible. In audience segmentation work, we regularly see high-completion, low-retention cohorts — users who moved through a funnel without a single point of engagement strong enough to encode brand preference. Smooth ≠ sticky. The flows with one deliberate moment of pause — a confirmation screen that asks a real question, a product recommendation that requires a single meaningful input — tend to produce users with materially higher 90-day retention rates.
This isn’t nostalgia for bad UX. It’s a call to design friction with intention rather than eliminating it by reflex.
AI Is Widening the Stakes of Every Design Decision
Smashing Magazine’s Andy Budd frames it clearly: AI is giving designers more autonomy than ever, faster. Fewer approval gates, faster iteration, more output. The bull case is obvious — designers can test and ship ideas that previously required cross-functional sign-off. The bear case is quieter but serious: when designers have more power to act, the gaps in their strategic thinking become harder to hide behind process.
For Southeast Asian brands, this lands with particular weight. Markets like Thailand, Vietnam, and the Philippines are seeing AI-assisted design proliferate across e-commerce platforms — Shopee sellers generating product pages, Lazada storefronts spinning up banner variants overnight. The velocity is real. But the quality signal is getting noisier. When every brand can produce frictionless, polished UI at scale, the differentiator shifts to judgment — knowing when to add texture, pause, or personality rather than just optimising for tap-through.
That judgment can’t be automated. It has to be designed in, deliberately.
The Craft Argument Nobody Is Making Loudly Enough
There’s a third thread worth pulling here, from an unexpected direction. Illustrator Makitoy — featured in It’s Nice That — builds tiny collages from daily observations, each one the product of repetition and patience. Stamp-sized, tactile, specific. The work is the opposite of scalable.
What does that have to do with UX? More than it seems. The creative discipline of slowing down to notice — of treating the mundane as raw material worth examining — is exactly what’s getting squeezed out of AI-assisted design workflows. The interfaces that generate the most engagement in Southeast Asian super-apps (think LINE’s sticker ecosystem, or Grab’s gamified loyalty mechanics) aren’t the frictionless ones. They’re the ones that feel made — that have a texture suggesting a human noticed something true about the user and encoded it.
For design teams, this is a practical prompt: build deliberate observation into your sprint cycle. Not user testing — that’s downstream. Actual observation of how your users live, what they pause at, what they share. That input can’t come from a prompt.
Implementing Intentional Friction Without Losing Conversion
The operational question is where to place friction without torching your funnel metrics. A few principles that hold up across markets:
Map moments of consequence, not just moments of drop-off. Standard funnel analysis finds where users leave. What it misses is where users decide — and those aren’t always the same screens. A decision moment is where a small pause could encode preference. Tag these separately in your analytics.
Test friction in confirmation, not acquisition. Adding steps to the top of a funnel is almost always wrong. Adding one meaningful touchpoint post-conversion — a personalisation question, a value confirmation — tends to increase LTV without hurting volume. Tokopedia’s post-purchase engagement flows use this logic: one prompt, one choice, and it feeds their recommendation engine while making the user feel seen.
Design for multilingual friction carefully. In markets with Thai, Bahasa, Vietnamese, and English running across the same interface, a friction moment that reads as thoughtful in one language can read as confusing in another. Test each language variant independently — character length alone changes the cognitive weight of a pause.
The broader question for 2026 is whether the design industry will treat AI-enabled velocity as permission to think less carefully, or as room to think more ambitiously. Frictionless was always a means to an end. The end was trust, preference, return. If smooth flows are now the floor — not the ceiling — then the differentiator has to be something AI can’t generate on demand: judgment about when to slow things down.
What does your product remember about your users — and do your users remember anything about your product?
At grzzly, we work with marketing and product teams across Southeast Asia to connect UX decisions to the audience data that should be driving them — mapping friction points, segmenting by behavioural signal, and building the decisioning logic that makes design choices defensible. If you’re rethinking how your digital experience holds up against AI-generated competition, Let’s talk.
Sources
Written by
Mellow GrizzlyTranslating raw data into activated audience segments, predictive models, and decisioning logic. Comfortable at the intersection of the data warehouse and the campaign manager.