Thanks to Scott & JP for reading early drafts of this text. Also huge thanks to Tim for our time at Kiwi as most of my insights come from that period.

***

I spent the last few years among crypto founders. I was interviewing them in my podcast, hung out with them at conferences, and became one while building an Ethereum Hacker News startup called Kiwi.

Since I left Kiwi in June, I had some time to step back and reflect. And I do think that consumer crypto is different than normal consumer startups.

And here's why.

We all know that startups look for Product-Market Fit, this magic combination of features and distribution that make them grow fast.

And looking for PMF is like searching for water with a magic rod.

The rod shakes when the water is near. But it’s easy to believe it shook, when it was just your trembling hand moving the rod, desperate for some water.

And in crypto this search gets even harder, because the environment is full of false signals that make your rod shake. So you can waste a lot of time going in one direction just to realize that there’s no water there.

Let's start with an example.

You probably heard of Farcaster. It’s a decentralized social app and this year they launched Farcaster Pro, which is kind of similar to Twitter Premium. During launch it didn’t offer many premium features yet.

But in 24h since launch they made over $1M in revenue. In consumer crypto! Seems like a moment to open champagne and dance, right?

But... when you look at the details, almost half of the people who bought this Farcaster Pro subscription had less than 100 followers. Which is weird given that Farcaster Pro was a power user product (that’s why it’s named ‘Pro’).

So why do you think these low-followers’ users decided to pay $120 for the subscription?

Some probably bought Pro to get a badge so their profile stands out and they can get some new followers. But many bought it to speculate. And it worked. First 10,000 users who bought Farcaster Pro for $120, ended up with $600 worth of airdrops the next day. 5X in 24 hours. Not bad.

Funnily this post above is not written by an airdrop farmer. It's written by Chris, a long-time Farcaster power user and early believer.

Which shows that at least two groups of people bought Farcaster Pro - speculators and believers. And this is first thing that distorts signal in crypto. When you're looking for PMF, you want to get market validation from people who just want to use your product. But - as we saw here - you can make $1M in 24 hours and still not be sure if you validated the market.

And this is not really Farcaster team’s fault. They want to build a genuine consumer product and the token that got airdropped hasn't been launched by them. But this is the environment where they - and all of us - operate in. Which makes it much harder to understand if you’re ‘building something people want’ as Paul Graham would put it.

And it’s just the tip of the iceberg.

When we launched Kiwi, you had to buy a $10 NFT called Kiwi Pass to fully use the product. In the first 3 months over 100 people bought it, including some high-profile users like Fred Wilson. It was 100% organic growth.

I remember thinking:
"We are getting paying users from Day 1 in the middle of the bear market? Aren't we onto something?"

But very soon we realized the story was more complicated.

We learned that some people bought the Pass because they were friends and wanted to support us. Some did it because they liked that we stood for decentralization and open source. And some paid these $10 because spending 'magic Internet money' feels easy, so why not give it a try?

This means that, even though we didn’t have airdrop hunters among our user base, only some % of our paying users actually wanted to use our product.

And that's pretty crazy if you think about it.

If you asked web2 founders, what are some vanity metrics they'd probably say 'Twitter followers', or 'page views'. But I don't think any would say 'revenue', because it’s the most fundamental startup metric.

Apparently, it's not that clear in crypto. You can see revenue growth and still not be sure if you’re heading in the right direction - either because of speculators or well-meaning supporters. So one of the best signs that you’re “on track to PMF” can lead you astray.

And the market structure is not helping.

There are about 45M monthly active addresses on Ethereum, if we count Mainnet and all L2s.

Even if we assume there's only one address per person - which is not true - it means that on a global scale Ethereum has less than 1% market penetration.

So - using classic Geoffrey Moore’s terms - most users (in the West, because it's different in Turkey or Argentina), are Innovators and Early Adopters.

These users are interested in the space so much that they will test your product even if the UX is bad. And that's nice, but this blade cuts both ways. Because these users are so open to testing new products, it also means that your usage chart may skyrocket in one week and next week your users could start looking for the next shiny thing.

So that's problem number one. These users are less sticky than “normal users”

Also, crypto market is very diverse.

At Kiwi our userbase was full of devs, founders, artists, VCs, and researchers, and their expectations of what content should we cover was diverse, too. Some wanted more technical content, others wanted more art, others wanted more essays. It was hard to make everyone happy without losing focus.

And this diversity is true for whole crypto, where you also have speculators, philosophers, cryptographers, TradFi people and so on. This means that this - already small market - is further divided into these subgroups, which are further fragmented across different L1s and L2s.

This is a challenge for growth (how do you even get these users if there are so few of them?) and for product design (how do you sort out feedback from multiple user groups who are here for different reasons and their requests are mutually exclusive?).

In big markets like coffee, you have space for many niches. There's even a multi-million dollar business run by veterans who want to serve coffee to people who love America.

But in crypto market it’s easy to pick a niche that’s too small and then, although people keep using your product, you can’t build a sustainable business because you just won’t find enough users to cover your costs. 

As pmarca said
“In a great market—a market with lots of real potential customers—the market pulls product out of the startup”.

But - outside stablecoins, speculation, and some niche use cases like prediction markets or collectibles, we don’t have ‘lots of real potential customers’ yet.

So on top of people who distort the signal, userbase that goes after ‘the next shiny thing’, we also have a very small market. And this is critical if you want to find PMF.

Most people think that PMF equals growth. 

App is suddenly getting 100k downloads in a week? PMF!
All your friends start using the product? PMF!
Protocol starts making $10M volume in a month? PMF!

But PMF is not only about growth. It’s also about retention.

If customers came to your shop, looked around and left, would it be a successful shop? Of course not. Yet we often think that the app getting many downloads already has PMF, forgetting about counter-examples like Clubhouse that died after its short-lived virality.

And there are many stories like that. When the Twitch team built the early versions of their app, they generated 16M downloads in 4 months. And - according to their founder - still didn’t have PMF.

If you want to hear the story, go to 7:30 in the video below:

And they didn't have PMF because their retention was so low, that their app was basically burning the market. People came to their shop, left and never returned.

We all know this high growth/low retention stories in crypto, because we all have seen airdrop farmers.

But it goes beyond that.

There are also waves that your product can surf. If you built an NFT marketplace in 2020, you got a lot of attention. Now try to build it in 2025, and see how it goes. These waves come and go, and you might believe that you made it, but then the meta changes and your product goes down with the wave.

If you'd like to dive deeper into retention, here's a good blogpost by Andrew Chen and a good video by Chamath where he explains why meteoric growth typically doesn't work.

There are also two other factors at play.

In crypto there are dozens of zombie projects which suck out retail’s attention and capital. Since it’s taboo to proclaim the project dead, founders ‘exit to community’ and their users stick around projects’ Discord channels believing their tokens, NFTs or points will one day pay off.

And this is is a brutal zero-sum game. If users still believe that Cardano is going to win, they won’t spend their time learning about Ethereum or your dapp because they are already in the propaganda content pipeline of their favorite network.

Another factor is that crypto token prices may impact your retention.

In 2024, during Base memecoin mania, we saw Kiwi usage going down. And why wouldn’t we? 

When someone can spend 15 minutes reading an essay about decentralization or 15 minutes finding the next 100X coin, many people - even those who are here for the right reasons - will choose the second option.

So to sum it up: we have speculators, users who support you but don’t use your product, small and diverse market, challenges with getting retention, and token prices making the attention game harder.

When I started writing this essay, I just wanted to describe the reality as I saw it. I didn’t mean to discourage founders from building in consumer crypto.

So I owe you some solutions. None of them turned Kiwi into the ‘next big thing’ (yet!) but they definitely made our lives easier.

Don’t add distortion

If you start a points, or referral program it will be harder to understand if people like your product or they just expect some money in return. So don't do it if you're looking for early signal. It’s a trap!

We once launched a referral program at Kiwi, but the results where underwhelming. Until one day, we saw a big influx of users. Once we started digging, it turned out that most of them were from Korea.

We quickly learned that one Korean influencer shared his referral link saying that people would get airdrops. We got many signups but few longterm users. And since we were looking for a signal and not extra $200, we just killed the program.

BTW not adding distortion is one of the reasons why our Kiwi Pass used to be non-transferrable - we didn't want people to speculate on it, as it would add more noise.

At Kiwi we rewarded users with karma. Like on Reddit, the more people upvote your posts, the more karma you got.

You could see the most active curators on leaderboard and we also credited them in our weekly newsletter, so they got some recognition in return for their work. It was a positive social Scooby Snack for being a good user, not a farmer.

We also had clear guidelines, explaining what type of content we are looking for. If someone shared low-effort spam, we could easily kick them out because their links didn't comply with the guidelines.

If we were in doubt, we checked onchain history to decide if they were a normal user who just didn’t read the guidelines or a farmer. You can’t add guidelines to every product, but there are many ways to reward good user behavior.

Talk to your users

When you minted a Kiwi Pass, we invited you to our Telegram groupchat. This is an amazing thing about crypto - most people use the same communicator app. And it's 10X easier to get casual feedback on Telegram than through e-mail .

So on Kiwi groupchat we shared product updates, discussed crypto news, gathered feedback to our feature ideas and so on. And it also gave us access to DMs of our most engaged users.

We spoke with a few hundreds of these users, either on Telegram or video chats, and learned what could we do to make the app more interesting. Sometimes I felt like a Jehovah witness asking if they wanted to speak about Jesus, but it was worth it.

If you get feedback, it’s critical to understand where it comes from.

There's this 90/9/1 rule of social apps that says that 90% of your users are just readers, 9% interact with the content and 1% users who post a lot. At Kiwi it was kind of similar.

We collected feedback from different types of users and then worked on features to fix our current bottleneck.

When we didn't have enough content, we built a Chrome extension for Creators that let you submit links with one click. When we didn’t have enough comments, we focused on features for Contributors like improving our comments editor. And when we wanted more readers, we focused on features for Lurkers, like apps' loading time, feed algorithm and so on.

Understanding which group of users you are talking to might help a lot.

When we saw someone becoming inactive, we reached out, and tried to understand what happened.

People often tried to be polite so they said they were “busy”, but we dug more, and some people became brutally honest.

Some said that they “Got logged out a few times and got tired of logging in again” or “Didn’t get enough upvotes for their links”, or just “Felt like reading our links is a bigger investment than scrolling the Twitter feed” and that was some information for us to work with.

Which Winnie the Pooh would you prefer to serve? Normal one or tuxedo one?

When I doubt, choose the tuxedo one, as he has more money to spend. You can have a few users but still make a lot of money if your Average Revenue Per User (ARPU) is high.

Last month on Ethereum Aave charged over $60M in fees with about 25k monthly users. That’s over $2,400 in fees per user per month. And that's because some whales borrow and lend millions of dollars. Of course not all fees go to Aave treasury but it's still a wild number.

So you don’t need many users, you need the right users.

Similarly, Blur focused on a very small niche of pro traders and also did great numbers when everyone thought NFTs are dead.

If your niche is too small but you believe it’s significant, you can also grow it or wait until it grows itself, like sensei from Ninja Turtles.

In 2022 privacy on Ethereum was a tiny niche. But in 2025 it is already big enough for Raligun to generate $5M in revenue. And since they’ve been around for 3 years, they became a top-of-mind solution when the market has grown. Same with OpenSea that built its presence before the NFTs became "the next big thing".

A more radical way to expand the market would be crossing the chasm and bringing normies in.

This is what NBA Top Shot did - their NBA partnership made collectibles appealing to average Joe and gave them an easy way to buy their first NFTs. And it worked!

Another example is Polymarket. They pushed prediction markets to mainstream - part of it is god-tier social media game, part of it is correctly predicting the US presidential elections, part of it is getting creative with their bets like.

Larry Page says he's looking for toothbrush products - something you use once or twice a day and it makes your life better.

If you focus on one core thing your product does and it can be done daily, it helps you in two ways.

First of all, it's obvious what's your product about. For Kiwi it was "finding great Ethereum content". And when we had great content in our feed ,we saw our usage numbers go up - because that's what people came for. It's also much easier to get feedback from the users when you do one thing - you have 20 conversations about one thing instead of 20 conversations about 20 different things.

Also, when you focus on daily problems, you get feedback daily. So you can correct your action more often than if you focused on monthly problems.

It doesn’t mean you can’t build a great product around weekly or monthly problems. Most people don’t stake their ETH everyday, and Lido is still very successful. But looking for daily interactions might help to get more signal.

If I prepare instant noodles, my Time To Value is about 1 minute - I just pour hot water and I get a meal. If I order something at the restaurant, my time to value is about 30 minutes, but the food is more tasty so the value I get is higher than with noodles.

In our case, our Value came from content. But the problem was that most of our content was pretty long, so we were like a restaurant - you had to spend a lot of time to get a very nice meal.

But we found out that on Hacker News many users don't click the links - they just read the discussion below which is way faster to "eat". That motivated us to make comments more popular - we added comment previews, emoji reactions, better way to write them on mobile and so on.

We also experimented with more instant noodles content like memes or charts, that might provide less value but you get it right away.

If your mom downloads your DeFi aggregator app, the Product-User Fit is low. She’s like a bird at the meeting. But if you onboard some DeFi veteran who makes 5 txs a day, the Fit is good. So the better ‘Product - User’ fit you get, the higher retention you will probably get.

When we went to Ethereum conferences, many people were interested in Kiwi and downloaded the app right away, and some became long-time users. That’s because ‘Product-User Fit’ was high, and that’s also why it costs $10k+ to get a sponsorship at top conferences - all products want this kind of userbase.

I know that Ethereum is full of desktop-first Harolds.

And yes, Figma and Cursor don’t need to care about a mobile app, but consumer apps need to. So please have a decent mobile app. We tried PWAs and they are fine for some time, but it's not good long-term because analytics is not good, push notificiations don't work as well, and it crashes more often because people use different mobile browsers.

So just get an iOS or Android app.

Andrew Chen has this concept called "Little channels". When your app has 100 users, and you get 10 more users in a week, you've just grown 10%, which is great.

So when your product is small, you can focus on smaller, high-signal channels like subreddits, forums, niche podcasts and so on, instead of trying to get viral on Twitter. This is how ZORA got their first users - they just DM'ed artists on Instagram.

When you're small you can even be like this guy on a meme, explaining what you do to people, one by one.

We went with Kiwi to specialized podcasts like Dev N Tell, became visible among Gitcoin projects, onboarded people one by one and so on. Some of these ideas didn't work, but some did well, like going to an ENS ecosystem call where 50% of attendees signed up to Kiwi right away.

Thanks to that laser-focus, you can find people who actually care about the tech more than the token price. Plus they will probably retain better.

Great thing about ETH conference is that you barely hear anyone talking about prices there. People are in it for the tech.

This is btw a pretty non-bullshit way to gather feedback.. This is how we learned what to improve in our onboarding and why mobile Ethereum wallets don't work. It was also useful to see how our app interacts on different browsers, operating systems, and wallets. 

This is btw how Railgun learned how to improve their Docs. 

So this was the last piece of advice from me.

All this advice is focused on early projects.

And by early I mean: ‘You just launched, and are looking for a signal to understand if you go in the right direction’. If you need to pump your numbers to raise the next round, or already have some PMF and just want to scale, it might be better to trade some noise to get more growth.

Speaking of early stage projects, my final point goes without saying, but I will say it anyway. All these ideas won’t work until you get your product out there. So it’s better to launch early.

When Tim launched Kiwi, you had to download Kiwi’s GitHub repo, set up a node and submit your link via CLI. It was - from the consumer product POV - overly complex. But it was enough for some users to get interested (including me), and this is how we started working with Tim. 

And looking for signal is tricky. The way I see it is that startup founder lives in a 24/7 casino, playing a game. In this game, if you put the right features and distribution pieces in the right order and click “Ship!”, you might hit the jackpot. But even for experienced founders and VCs it’s hard to tell if the combination is right before you click the “Ship!” button.

It's more like arts than science. In arts you can build your taste and feeling for the trends, but it doesn't mean you can always predict which song will become the #1 Billboard hit.

Black Sabbath’s biggest hit - "Paranoid" - which they choose to end their band's last performance was an album filler written in 20 minutes. Quentin Tarantino and his director friends thought that “Pulp Fiction” was going to bomb. “Shawshank Redemption”, the #1 movie of all time on IMDB did - in fact - bomb before it became a video rental hit and built its legacy.

It's similar with startups - it's hard to predict what's going to work. In the end, consumer products and consumer culture have lots in common.

So please don't get discouraged by all challenges I mentioned, get your product out there, and good luck!

If you liked this essay you can subscribe to my blog - I typically post once in a few months,
so I won't spam you: