Designed Friction

Designed friction is the intentional addition of resistance in user interfaces at moments where an accidental action carries real consequences. Top teams stopped treating every extra click as an enemy and started using friction as a material that communicates care and prevents expensive mistakes. The decade long obsession with removing every possible tap produced products that optimized for engagement at the cost of trust. In 2026 the smartest designers map their flows for irreversible actions, money movements, and trust sensitive moments then tune the exact shape of confirmation to match the risk instead of defaulting to zero. A product that respects its users builds in the pause that lets them catch their own errors before they happen instead of offering instant undos after the damage is done. The second confirmation on Apple Pay and the typed repo name on GitHub both prove that well placed friction reads as respect rather than roadblock. Frictionless was never the goal. Effortless was.

This approach has nothing to do with the accidental friction that comes from poorly optimized code, unclear copy, or forms that collect unnecessary information like a fax number in an age of digital signatures. It also differs completely from dark patterns that use friction to trap users in subscriptions or hide the delete account button behind multiple confusing screens and guilt inducing language. Those tactics protect company metrics at the expense of user goodwill and eventually destroy brand reputation as seen with several high profile apps in 2024 and 2025 that faced mass negative reviews and app store backlash after hostile cancellation flows. Designed friction never trains users to ignore prompts by overusing them. It never sits at the beginning of the journey before value has been delivered. It never applies uniform weight to actions that carry wildly different stakes from archiving a document to deleting an entire company workspace. Teams that confuse these concepts end up with products that feel both sluggish and careless at the same time which is the exact opposite of the respect designed friction aims to convey.

GitHub provides the canonical example with its repository deletion flow. Users must type the exact name of the repository before the button enables which cuts straight through autopilot behavior that would otherwise delete months of code with one distracted tap. Cursor took this lesson into the AI space by forcing developers to review and accept every suggested code change through a detailed diff interface before any edit reaches the file system with many teams in 2025 learning the hard way that skipping this step led to corrupted codebases and lost trust. Teams that launched autonomous coding agents without this approval layer watched their users flee after the AI deleted critical files or sent test emails to entire customer lists. Stripe implements designed friction in its checkout by splitting the process into two distinct screens so buyers see the full summary including merchant name amount and what they receive before money leaves their account. Linear built one of the most elegant versions across an entire product where archiving a project requires one click since the action reverses easily while deleting a project triggers multiple escalating confirmations including a full list of dependent items that will be lost forever. Notion follows the same graduated approach where permanent deletion demands typed confirmation but simpler archiving does not. Apple Pay uses hardware level friction with the double side button press that gives users a full second to change their mind on transactions. Gmail added an undo send window that lets users retract emails for 30 seconds after hitting send which handles the publishing moment without blocking the initial action. Slack warns users before broadcasting to large channels with a clear message about how many people will see the message. X formerly known as Twitter added confirmation prompts on replies that match certain emotional trigger words to prevent career ending posts. Claude requires approval before artifacts can execute code or make external calls. Webflow shows a full publish checklist before any site change goes live to the public. AWS makes resource termination require both a confirmation modal and typed verification for critical infrastructure. These real world implementations from 2023 through 2026 demonstrate that friction comes in many shapes and the best teams pick the right one for each context rather than defaulting to the same modal everywhere or removing all pauses in the name of speed.

Use designed friction on irreversible actions like dropping a production database on AWS where they force you to type DELETE in capital letters or force pushing over production code in Git. Deploy it on every money flow because the rise of one click subscriptions created the most complained about category in every app store for good reason with users feeling robbed rather than served. Insert it before any AI tool call that touches files sends messages or spends money since every such action is effectively a money flow where the currency is user trust and burned trust cannot be refunded. Apply it to account deletions and destructive settings so users never discover at 11pm on a Tuesday that they permanently removed the wrong workspace with no recovery path. Add it to sharing and publishing moments because the cost of sending the wrong draft to a client in Figma or posting an unfinished blog to your entire mailing list can damage careers and relationships in ways that no apology can fix. Place it at points of sensitive data access like updating recovery emails or approving large transfers or new device logins rather than on every single balance check. The six shapes range from a light 200 millisecond pause on low risk buttons through simple modals double confirms type to confirm requirements cooldown timers and finally undo windows like Gmails send undo that give grace after the fact. Avoid it completely at the front door where extra steps before users experience value increase churn rates as seen in numerous failed onboarding flows from 2023 to 2025 that never let users try the product before demanding full profile completion. Never deploy it as dark pattern retention like the fitness apps that required phone calls to cancel or buried cancel buttons under guilt modals that made users feel ashamed for wanting to leave. Skip it during initial onboarding because users have not yet committed and will simply leave instead of fighting through setup wizards that hide the core value behind preferences. The correct pattern is progressive disclosure where the product earns more information over time as it proves its worth. The rule is simple. Place friction at the action not at the entrance. Scale it exactly to the cost of a mistake. Make sure it reads as care rather than obstruction. Teams that follow this approach run every flow through an audit that asks whether the action can be undone what exactly the user loses if they tap by accident where the pause actually lands in the journey and whether the user will feel respected or annoyed by the interruption. Get the answers correct and the product earns lasting trust while turning high stakes moments into opportunities to demonstrate reliability. Get them wrong and even the best visual design cannot save the experience from feeling either reckless or hostile.

Designed friction turns good products into products that users trust with their most important work.