Dead ends are the silent killers of user flow
Users don’t bounce because of one bad screen. They bounce because you didn’t tell them what to do next.
Every screen in your product is a small conversation. Good screens end the conversation with an obvious ‘and next, do this.’ Bad screens end in silence. The user is left holding a half-finished thought and an awkward question: now what?
What a dead end looks like
A confirmation screen that says ‘Thanks!’ and offers nothing to do next. A search results page with zero results and no suggestion. A success flow that hands the user back to a generic dashboard. An onboarding step that completes and… just sits there.
None of these are broken. They all ‘work.’ They just don’t complete the loop. The user has to guess where to go next, and guessing is the last thing a user wants to do when they’re already uncertain.
“Every screen ends in one of two things: a next step you designed, or a next step the user invents. Usually the invented one is ‘leave.’”
The three nexts
Every screen should offer at least one — ideally no more than three — explicit next steps. The best products pick:
The obvious continuation. What 80% of users will want to do. Make this the biggest, most prominent option.
The power-user extension. What a 15% user will want. Secondary button.
The escape. What the remaining 5% needs — to back out, rethink, try something else. A quiet link, not a shout.
If you’ve named three, you’ve probably captured everyone. More than three starts to look like you don’t know which one matters.
Dead ends in unexpected places
The biggest dead ends are in success states. After a user finishes onboarding. After they complete a purchase. After they invite a teammate. These feel like endings, so we treat them as endings — but each of these is a high-trust moment where the user is most likely to accept another suggestion from you.
Missing this moment is the single biggest wasted opportunity in most products.
What good flows feel like
They pull the user forward. Every screen suggests the next one. The user doesn’t consciously notice the suggestion — they just keep moving, and twenty minutes later they’ve done more than they planned to.
Bad flows stall. Good flows hum.
Dead ends are rarely a design problem — they’re a thinking problem. The fix is almost always just deciding what should happen next, and saying so.