Ask a room full of users what they want next and you will get a remarkably consistent, remarkably unoriginal list. Faster load times. A better search bar. Dark mode, again. None of these answers are wrong, and none of them are useless, but they share one quiet, structural limitation that most product teams never name out loud, every answer on that list is something the user has already seen somewhere else. A vote is not really a measure of what people want most. It's a measure of what people have already been shown is possible, ranked by how much they liked it.
This is the actual reason feature requests cluster so heavily around table stakes. Nobody in a user interview is going to ask for a capability that doesn't exist anywhere in their frame of reference, because you cannot want, in any specific, actionable way, something you have no vocabulary or mental model for yet. The most requested features in any product are, almost definitionally, features a competitor already shipped somewhere, or a pattern the user has learned from a completely different app and is now asking you to import. That's not a criticism of users, it's just how imagination works, it operates by recombining things you've already seen, not by inventing things that have never existed anywhere.
This is exactly why building only what users explicitly request tends to produce a product that converges toward its competitors rather than differentiates from them. If every company in a category is running the same feedback process, surveys, interviews, upvote boards, and every user pulling from the same pool of things they've seen elsewhere, then every company ends up building the same requested features in roughly the same order, because the requests themselves were never actually generated by any one company's unique context, they were generated by whatever the entire category had already normalized. The result is an entire market of products that all look increasingly alike, each one dutifully checking off the same list of table stakes their users kept asking for.
The features that actually separate a product from its category tend to arrive from somewhere else entirely, not a vote, but a team noticing a friction users hadn't yet learned to name, and building a solution before there was language for the problem. Nobody was requesting a swipe gesture before touchscreens existed to swipe on. Nobody was upvoting a request for a recommendation algorithm that could predict what they'd want to watch next before recommendation engines existed anywhere for them to have experienced one and formed an opinion about it. These weren't things that lost a popularity contest to more requested features, they were never eligible to be voted on in the first place, because the option didn't exist yet in anyone's head to choose from.
This creates a genuinely uncomfortable position for any team that leans heavily on customer feedback as its primary source of roadmap direction, because the process that's supposed to reduce risk, ask the users, build what they ask for, systematically filters out exactly the category of idea most likely to create real differentiation. A feedback board full of upvotes will always favor the familiar ask over the genuinely new one, because familiar asks are easy to understand and immediately upvote, while a genuinely novel idea usually requires actually experiencing it before anyone can tell whether they'd want it, and you cannot experience something that hasn't been built yet in order to vote for it.
None of this is an argument against listening to users, ignoring feedback entirely produces its own kind of failure, usually a product so disconnected from real usage that it solves problems nobody actually has. The point is narrower and more specific, feedback and voting mechanisms are excellent tools for prioritizing among things that already exist in some form, competitor features, known patterns, established categories. They are close to useless for surfacing the next genuinely new thing, because voting requires options, and the option that would actually move the product forward the most usually hasn't been invented yet by anyone the user has ever interacted with.
The actual job, underneath all the feedback boards and upvote counts, is building the thing users would vote for if it existed, before it exists, using judgment rather than a tally, because by the time enough people can articulate wanting it clearly enough to vote for it, every competitor watching the same feedback board already knows it too.
August 20, 2026
You Can't Vote for What Doesn't Exist Yet
The loudest, most requested features are usually the most obvious ones, which means every competitor is already building them too. The features that actually create an edge are almost never the ones users asked for, because nobody can request a solution to a problem they don't know is solvable yet. Voting only works on options that already exist in someone's imagination.