mar 2026
there is a pattern in every technology transition and it is slightly humiliating once you notice it.
the companies that end up owning the mainstream do not describe themselves using the language of the technology they are built on. they describe the thing the person wanted.
none of those companies were being coy. they had just worked out that the substrate is not the product, and that the number of people who care about the substrate is always, always smaller than the number of people who want the outcome.
and the tell that a technology has genuinely crossed over is that it goes invisible. you used satellite positioning four times today. you did not think about satellites. most people using cloud infrastructure could not define cloud infrastructure. the technology stops being a thing you talk about and becomes the floor you stand on.
this is where i think the usual version of this argument stops too early, because it treats it as a marketing observation. rename the category, soften the jargon, talk about outcomes.
that is not it. the label is a symptom. what it is a symptom of is that the product still requires the user to understand the machinery, and that requirement is real, and no amount of homepage copy dissolves it.
here is the actual thing.
seven steps. six of them are pure cost to the user, and every single one is a place they can leave, and you paid money to put them at the top of that list.
and i want to be precise about why this is fatal rather than merely annoying, because “onboarding is hard” is the most ignorable sentence in product. it is fatal because of what it does to your economics. if a user only becomes valuable after seven steps, and most of them do not finish seven steps, then your cost of acquisition is not what you paid for the click. it is what you paid for the click divided by the fraction who made it through. and then you have to pay again, because someone who never reached value has no reason to come back on their own.
a product that has to be paid for twice has not been made easy. it has been made expensive.
the products that work deliver something before they ask for anything. sign in with a thing the person already has, then hand them the outcome, ideally before their brain has finished registering that it did any work. after that you can ask for a password, an email, a wallet, whatever you need. you have earned the right to ask, because you already gave.
the bad version is the one everyone builds, and it is bad for a reason that feels responsible at the time: email, password, verify, configure, choose your plan, and then a blank screen with a tour. every one of those steps was added by someone sensible solving a real problem. collectively they form a wall.
so make it invisible. hide the keys, abstract the chain, sponsor the gas, log them in with google, and let them never learn the word bridge.
and this is where i think the “just make it invisible” version of this argument becomes genuinely dangerous, because there is a floor and people keep going straight through it.
some of what looks like friction is not friction. it is the product. self custody is not an implementation detail you can tastefully conceal. if the user cannot tell whether you can freeze their money, you have not simplified anything, you have rebuilt a bank with extra steps, and the entire reason the technology exists was to not be that. we have watched this happen at scale and repeatedly, and every time the interface was lovely.
so the rule i would actually write down is narrower than “make it invisible”:
hide the mechanism. never hide the guarantee.
nobody needs to know what a bridge is, what chain they are on, what gas costs, what an approval transaction does, or that any of it exists. all of that is machinery and machinery should disappear.
but whether you can be locked out, whether someone else can move your funds, whether this works if the company dies, whether the thing is actually yours. those are not machinery. those are the product, they are the reason to pick you over the incumbent, and a user who cannot answer them has been handed a worse version of what they already had.
the invisible payment i wrote about a couple of weeks ago is a good illustration of both halves. the machinery is entirely gone, which is why it works. and the guarantee is entirely visible, if you look: one party can see it and one party can stop it. the user knows exactly what they have. it is just not what a crypto user would want.
i do not think this is free, and the version of this take that presents it as obviously correct is skipping something.
the label does real work right now. it gets you into the right rooms, it hires people who already believe, it raises money from funds with a mandate, and it reaches the existing audience who are the only people currently transacting. drop it and you give up distribution you have today for a market you do not have yet.
so it is a timing call, not a principle. leading with the technology is correct while the market that cares about the technology is bigger than the market that cares about the outcome. that is true early and it stops being true at some point, and the mistake is not choosing wrong at the start. it is failing to notice when it flipped, because the label is comfortable and it is full of people who agree with you.
it takes about two minutes. open your homepage, delete every crypto specific word, and read what is left.
if there is still a clear product there, you are fine and this is a copy exercise. if what remains does not make sense to somebody who has never held a token, the words were never the problem. the words were load bearing, which means the product is the machinery, and that is a much longer conversation than a rewrite.
my guess is the biggest company in this space in five years will not have the word chain or crypto anywhere on the front page. not because they are hiding it. because by then it will be as interesting to mention as which database they use.