> ## Content Index > Fetch the complete content index at: https://www.web3-advertising.com/llms.txt > Use this file to discover other available public pages before exploring further. # Every startup does marketing. Most do it by accident. - URL: https://www.web3-advertising.com/every-startup-does-marketing-most-do-it-by-accident/ - Published: 2026-08-07T11:05:04.000Z - Updated: 2026-08-07T11:05:04.000Z - Description: Founders think marketing starts when there is budget for it. By then the headline is written, the channel is picked, and nobody wrote down who any of it was for. The cost is not a bad website. It is not being able to tell what went wrong. - Author: Filip Josipović The product is live. The site works. Nobody is coming. So what is wrong? Is the message off, is the channel wrong, or is the product itself not solving anything anyone cares about? If you never wrote down who this was for and what you were claiming, that question has no answer. All three failures look identical from the inside. You will guess, and then you will spend six months or a year chasing something you were never sure you should be chasing. At the end of it you will either decide you need a rebrand and rebuild everything, or conclude the product does not serve anyone and walk away from something that was fine. Most founders come out of that stretch with one conclusion: we should have started marketing sooner. That is where the thinking usually stops, and it is the wrong conclusion. You were already doing marketing, and nobody planned any of it. Somebody wrote the headline on the site. Somebody picked the first channel. Somebody decided which feature goes at the top of the deck, and chose the three words describing the product in an investor update or in a reply to a stranger who asked what you are working on. Every one of those is a decision about who this is for and what they care about. Your marketing started the first time anyone outside the team learned the product exists, and by launch day most of it had already happened. There is no shortage of advice telling founders to define their customer earlier. What it misses is that the definition already exists, several versions of it, none of them written down. So there was never a choice between doing marketing and not doing it. The choice was between making those decisions once, on purpose, or repeatedly, by accident. Accidental marketing does not cost you a bad website. It costs you control over what your product is taken to be. Everyone who touches it forms their own version, the market picks up whichever version it heard most often, and you never find out which one that was. 0:00 /33:51 1× ## What accidental marketing looks like Most startups today, especially in web3, are tech startups founded by tech people. Everything worth caring about is the code and the product, and the thing that gets the least attention is anything to do with marketing. Early on that is the right order. AI has made this worse. Almost every startup I look at now has a site that reads well at first glance, which not long ago was a fair signal that someone had thought about it. Go one level deeper and it often does not hold together. Either it does not quite make sense, or it makes sense and could be far better. None of it is incorrect. It just has no depth, nothing you could read and come away knowing what the company is or what the service actually does. There is a version of this that is specific to crypto. In the good case, a few developers notice a real problem, know who has it, and build a solution. Those teams explain themselves easily. But in the majority of cases startups get built where the money is at that moment. A couple of years ago that was NFTs, then lending protocols, and lately it looks to me like RWAs. If you launch into a crowded category with no clear statement of why anyone should pick you, your communication will stay vague no matter how well it is written. The product ships, and the first thing the market learns about it was decided by whoever happened to be writing that week. ## The comparison nobody named If you fix one thing, fix this one, because every other decision hangs off it. **What would your customer use if you did not exist?** Most teams never ask it. The ones that do answer with competitors, the other funded companies in the category. That is almost never the only answer. You do have competitors and they matter, but the thing you are actually being measured against is usually one of these: - A manual process somebody repeats every week. - A spreadsheet. - An older tool that is worse but already installed and already approved. - Leaving the problem alone. Here is the mechanism, in a made up example: say you built something that watches positions across several protocols and warns you when one goes wrong. - **Frame it as a portfolio tracker** and you are competing with every dashboard in the category, where the comparison runs on how many chains you support and your alerting is a line item. - **Frame it as a risk monitor** and the comparison runs on how early the warning arrives and how often it is wrong, which makes your alerting the whole point and chain coverage a footnote. Same code, but two different products as far as the buyer is concerned, and only one of them is a fight you can win. Choosing between frames comes down to one question: which one puts the thing you are best at in the middle of the comparison? Now put the wrong frame on the site. A buyer arrives comparing you to the spreadsheet they check every morning, and your homepage is busy explaining why you beat a named competitor. None of it connects. The copy is fine, but you answered a comparison nobody was making, and you will read that failure as a conversion problem. ## What deliberate looks like This is what brand discovery actually is, and it is worth saying plainly because the term suggests something far heavier than it needs to be. It is one short document, not a brand system, holding three more decisions after the one above. **Who has this problem, and what does solving it give them.** The alternative narrows the audience for you: whoever is currently tolerating that workaround is your starting list, and what it costs them is the value you are selling against. **What we say to those people.** The claim, who it is aimed at, and what backs it up. If the claim is that setup takes minutes, the proof is a real user who did it in minutes, not a figure someone estimated in a meeting. Early on you will have claims with nothing behind them yet, and that is normal. Mark those as unproven in your own document, which nobody outside the team ever sees, so you know which proof to go and get rather than inventing a number for the site. **Where those people already are.** Channel is the last decision, not the first. If the people you want found their current workaround through someone they trust rather than through a search, that already tells you something about where to show up. Testing it is less formal than it sounds: a Discord you actually talk in, a subreddit where the problem already gets discussed, or ten people you know in the industry will tell you more in a week than a survey will. If you have been around long enough to build the thing, you probably know enough people to check whether the claim lands. All of it is a first draft, and it helps to write that on the document, because the draft is the point. You are not trying to be right in month one. You are trying to have something specific enough to be proven wrong quickly, which is exactly what accidental marketing can never give you. ## Who is this for, and what actually makes them move When people hear personas they think age, location, job title. There is a meme comparing the demographic profile of Ozzy Osbourne and Prince Charles. Same age, same country, same income bracket, same marital status. Two completely different people. That is why a persona built only on demographics is useless. It describes who somebody is and never explains why they act. When someone has to actively choose a tool or a service, and usually justify that choice to somebody else, a better frame is that they are not buying a product. They are hiring it to get a job done. That job has three parts: - The task itself. - How the person wants to feel while doing it. - How they want to be seen by their team or their boss. Teams write down the first and stop. In B2B the third one often decides the purchase. Then there is the question of what makes someone switch at all. Four forces decide it, two pushing toward you and two holding them in place: - Their situation is bad enough to act on. - Your option looks better than what they have now. - Anxiety about moving to something unfamiliar. - Habit, meaning attachment to the current way even when it is objectively poor. Accidental marketing almost always works on exactly one of those four. It makes the product look appealing. Anxiety and habit are frequently the stronger pair, and nothing in the message touches either, so the message reads well and nobody moves. For most early companies the real competitor is not another company. It is the way things are already being done, which is familiar, already paid for, and requires nobody to make a decision. The last piece is timing. Everyone in your target group has the problem more or less permanently, but very few are looking for a solution at any given moment. Something specific has to start the search, usually a new hire, a missed deadline, a budget review, or a question from someone senior that nobody could answer. Without the trigger you spend money reaching people who are not in a position to act. ## The internal cost nobody budgets for Accidental marketing has a second bill, and it has nothing to do with advertising. You have five or six people working on the same product and every one of them explains it differently. None of them is wrong, and each version is defensible on its own. Think about how conversations work. Nobody remembers a five minute conversation word for word. People remember the two or three things you emphasised. When a team emphasises the same three things, the market gradually forms a clear idea of what you do. Spread fifteen things across the same conversations and nothing forms. This is why the discovery work is worth doing even with no marketing budget and no intention of running ads. The output is internal. You never have to publish it. What it does is give four founders who have honestly never sat down and asked what are we in essence a shared answer, so everything built afterwards connects back to the same thing. Even when you turn out to be wrong, you will be wrong in one direction rather than five, which is a correction instead of a rebuild. The cheapest part of this is a short list of the claims you make about the product and the words you use to make them, along with the ones you avoid. It sounds trivial. It is the difference between four people writing in one voice and four people writing in four. ## Why you cannot trust your own read on this Three things get in the way, and none of them can be solved by trying harder. **You are not your customer.** Even if you were at the start, after a year of building you are not anymore. You cannot see the product with the same eyes. That is what happens to anyone who works on one thing long enough, and it is the reason the first version has to be tested rather than trusted. Frequently your users are hiring the product for something slightly to the side of what you built, and that is information, not a problem. **AI will produce something that looks right.** Anyone can now build a working app without writing a line of code, and it will run. What you cannot do, having never built one yourself, is tell whether it is actually sound or whether something is sitting behind it that will break later. You never made the thing, so you have no basis for judging it. Ask any current model for a positioning statement and you are in the same position. It reads well, and whether it is correct is a separate question that you can only answer if you know what you are looking for. **Research is hard here, and you still have to do it.** Many of the people who matter in crypto are anonymous and intend to stay that way, so there is no clean way to go and study them. You end up drawing conclusions yourself, or finding someone who was around at the time and asking. That difficulty is an argument for doing less of it, not none, because a single piece of information can move your whole focus. For a stretch there was a new protocol every other month handing out points ahead of a possible airdrop, and there is a whole community that does only that. I once spoke with someone running a large community in Southeast Asia who told me she did not know a single person on Ethereum at the time. Her people moved chain to chain, put small amounts into new protocols, farmed the points, and left. One conversation, and the picture of who was actually out there looked different. ## Whose feedback is actually worth anything Which brings up whose feedback is worth anything at all. You will not get useful product feedback from people whose motivation has nothing to do with what you built, and this is the trap that catches good teams, because the responses arrive and they look like data. The same problem shows up in any incentivised feedback. Post a form offering a reward for testing your tool and you will get volume from people optimising for the reward. A handful of real users found through word of mouth, properly compensated, is worth more than a thousand tainted answers. The tainted version looks better in a report, which is exactly why it is tempting. [How to run a user survey that actually qualifiesWe built our first product without asking a single user what they wanted. The survey that fixed that mistake is worth copying.![](https://storage.ghost.io/c/dd/d7/ddd7acdd-b0e2-462c-8132-53172eb6962a/content/images/icon/w3a-c28bc189-079e-46f1-9006-80e652235546.png)Web3-AdvertisingWayne Hattingh![](https://storage.ghost.io/c/dd/d7/ddd7acdd-b0e2-462c-8132-53172eb6962a/content/images/thumbnail/talk-to-your-users-2-08d6b4a9-6fde-4421-a4f0-29240fc9e029.png)](https://www.web3-advertising.com/how-to-run-a-user-survey-that-actually-qualifies/) Incentives are not the problem. If you are raising a pre seed and need to show that people use the thing, that campaign does its job. Hyperliquid is the example people reach for here, and I think they are right to. It ran points and incentives heavily, and the protocol also worked. The incentives brought people in, but the product is why they stayed. All of which is why the honest answer to most marketing questions is that it depends. Which channel, which message, which users to listen to, none of it has a fixed answer. A written position does not change that. What it does is turn that into something you can work with. If you have written down who the buyer is, what they would use instead, and what makes them start looking, then when a channel underperforms you have three specific things to check rather than an open field. Without that, it depends really means it depends on something, and nobody can tell you what. ## If you do nothing else You do not need to hire anyone to start. Sit the team down and answer three questions. 1. What does your customer do right now instead of buying from you? 2. What specific event would make them stop doing that? 3. Looking at the alternatives that already exist, what makes someone choose one over another? Write those down, check which of them apply to what you built, and those are your entry points. The first question is the comparison from earlier in plainer words, and it is the one to answer if you only ever answer one. You cannot skip marketing. It happens with you or without you. The only question is whether you find out what those decisions were worth in a fortnight, from a draft you deliberately set out to break, or a year from now, from a product you have started to believe nobody wants. --- *Filip Josipovic is a marketing consultant working with early stage startups.*