Original compilation: Block unicorn

Original compilation: Block unicorn

Acknowledging all the stupid things that are done in the digital asset space, and why we are doing it. It's also my attempt to determine where the industry is headed next, and it resonates with many of you.

With that said, I've laid out some frameworks for what might be next for startups and VCs in the industry. It's based on a mental model I use to think about which businesses are likely to take off at the bottom of a bear market, and I want to spend time working on it over the next few quarters.

In March 2022, I wrote about Aggregation Theory and its application to Web3. This will make it possible for a new generation of businesses to build and scale at previously impossible speeds. Think about OpenSea. No team verifies that every Boring Ape sold on the platform is an original. As long as the correct smart contracts are involved, there will be no concerns of duplicate asset transactions.

In traditional marketplaces, like Amazon, or Facebook's ad marketplace, the marginal cost of each new marketplace interaction rises as more users come to the platform because the potential for fraud is high. Even in digital consumer goods, like streaming music (on Spotify) or selling in-game items through Steam, the marginal cost increases with the number of users. When a platform has 10,000 users, the fraud checks you do are significantly different than what a platform with 1 million users requires, and blockchain tends to squeeze that cost down to zero.

The chart above shows the cost of a typical internet-native platform over the past decade. First, there is an initial cost accumulation in recruiting, developing and acquiring users. Then, as more and more users are added to existing programs, costs level off. Finally, once a critical mass for maintaining relative market share is reached, cash burn accelerates.

For a Web3 platform like Uniswap, the initial cost is in code deployment and auditing. Users bear the cost of adding liquidity to the pool. Therefore, there is no increase in marginal cost as the platform expands. Blockchain-related dApps are unique in that marginal cost (vs. venture capital) tends to remain flat no matter how many million transactions occur.

However, when building blockchain-first applications, there is a risk. That is liquidity. Most start-ups compete for one of two resources. One is attention. And the other is capital. Traditional web2 media platforms hack into your head or wallet through powerful network effects. Once their user base reaches critical mass, people are effectively forced to sign up to platforms like Facebook or WhatsApp because the price of not engaging is missing out on important updates or events. These platforms became the town square, where everything that happened in the world could be seen, the gateway to civilization (and the eventual collapse of civilization).

But companies like Meta are able to build this fluidity of attention because users can easily log into them. (I use the term "fluidity" to refer broadly to the amount of time a user spends on an app). WhatsApp, for example, hardly requires people to have an email address to join. This plays a key role in emerging markets like India, where phone numbers are the only form of digital identity users have.

first level title

public issue

Compared to today's Web3 applications, it's a complex maze of switching chains, wallets, and web interfaces before users get what they want. This is acceptable if you're targeting a niche group that is willing to suffer for the lack of alternatives. For example, stablecoins offer a better user experience in emerging markets than banks, which are too slow to process.

But if you want to scale to tens of millions of users, the situation is different. The challenge facing most dapps today is that they are competing for stock share in a decentralized ecosystem where the total number of Web3 users is currently less than 10% of the traditional Internet ecosystem.

The researchers found that today's dApps only attract users interested in volatility and risk. This may indeed be true, but this reflexivity works in two ways. 1) When the price is trending up, people flood into the same trade. 2) When the market crashes, people lose money and leave the industry entirely, overly volatile consumer facing apps that don't have the consumer's best interest in mind are not sustainable in a bear market. The speculative frenzy shoeing these products is a temporary hack that enriches random token holders.

Ignoring how easy it is to flee capital in the context of digital assets is the death knell for dApps. For scale, Binance had $3 billion out of its balance in 24 hours (Nansen Dec. 14). A bear market like the one we're in right now can kill startups through consumer apathy long before hacks or lack of venture capital.

first level title

fight against apathy

The market is a pendulum that swings between periods of excitement and periods of apathy. Bull and bear market cycles are a quantitative measure of human sentiment. But there is also a different way market cycles manifest: between periods of aggregation and unbundling, both of which are hangovers of fluctuations in consumer attention, let me explain.

Aggregation theory believes that bringing together a large user base can reduce sales channel costs and make new business models like OpenSea possible. Splitting is the exact opposite of what aggregation theory is about. During this period, founders instead focus on niche verticals with sticky user bases, rather than obsessing over growing user numbers during periods of market indifference.

Distribution costs increase when you have a small user base. That means startups need to be able to squeeze more money out of each user if they hope to survive a bear market. (CAC and LTV in Silicon Valley parlance). The way to make sure each user increases your revenue is to strip out all"good"The part that focuses on satisfying existing users, obsessed with those few users who are more willing to pay.

The growth of the internet has several examples of this, Amazon upended commerce as we know it, but it started as an obsession with helping nerds buy books online. What about Facebook? But its original example was getting Ivy League students to rate their fellow students, and of course it can be used today to overthrow governments and spread propaganda. What about Netflix? Their production budget was bigger than the education budget in India today, but they started off with a focus on delivering DVDs to homes. Do you know what I'm talking about?

There is a reason for this, products that focus on a single pain point stand out faster. When a company only focuses on a small subset of features, it reduces cash burn. And when something is simple, customer acquisition happens organically. These are the stages of rapid iteration of the product. As time goes by, because the product is constantly improving, users have proposed which products they want to add to the product."nice one"。

This has been shown in the past few bear market cycles. In their very beginnings, Nansen focused on allowing users to track wallets and token transfers; they have since expanded into DAOs, NFTs, and the multi-chain world. Binance was originally a place for spot trading. Since then, they have expanded to the industry's ecological economy, supporting derivatives trading and then OTC, games, VC and other businesses. In the end, CoinMarketcap and CoinGecko started out as platforms providing token price data, today you open CoinGecko and you see how they have expanded in due course to be a trading terminal, a portfolio manager, a research center and an API provider . Do you understand the trend change?

A handful of DeFi businesses have adopted the same strategy. For example, Curve started as an automated market maker focused exclusively on stablecoins. This AMM-based exchange has become one of the best places for millions worth of stablecoins to trade with each other, with low transaction slippage even surpassing centralized exchanges like FTX. By focusing on one specific vertical, they avoid going head-to-head with Uniswap, which is expanding into the long tail of unlisted assets.

But will these businesses (or dApps) remain aggregated? it's not true. Once a sufficient amount of liquidity—whether it’s attention or capital—comes into the product, the team may add other sub-features to better retain users within its ecosystem. In the case of Aave, they started their lending product very early, when there were maybe five credible players pursuing the opportunity. They have since integrated asset trading functionality within the AAVE interface, and by August 2022, they plan to release their stablecoin on the market.

first level title

The role of split leverage

None of what I've said so far is a fully tested theory. It’s common sense that founders should lower their CAC (customer acquisition cost) and focus on building tools that matter. All I've done so far is give a mental model and give examples of how splits have worked in the past. But if you think, I should just focus on splitting what I'm building since I don't have any PMF (data on user requirements), how would you go about doing that?

Founders can create a sub-functional product by focusing on implementing the key functions of the blockchain. If you are a founder, determine if there are other sub-functions that create other sub-functions in your product main function, because they may help you build a moat in the sea of ​​replication. The role of these levers is closely related to what I said before-first level title

capital efficiency

I believe that in the context of capital-intensive startups such as DeFi, product disassembly can be achieved by making the product more capital efficient. The basic argument is that consumers will go to the market with the lowest cost. If you can keep iterating on a feature to keep costs down, it will attract users.

I use "capital efficiency" as a metric - a measure of how much capital it takes a business to scale without bottlenecks (TVL). If $100 million in sales requires $1 billion in user deposits, then you have a debt machine. Providing incentives in the form of tokens can collect funds from users, but it means that the business may be exposed to hacking attacks on smart contracts.

There are some examples of this in the industry, for example, products like 0x Matcha (aggregation DEX) and Hashflow (l is a MEV-proof, zero-slippage DEX) focus on Protect consumers from MEV (miner extractable value). Therefore, they can offer better quotes than Binance for large orders converting ETH to USDC, despite the risk of centralization.

Hashflow has cumulatively completed nearly $12 billion in transactions by being the best place to convert a limited number of assets. In contrast, Uniswap pursues long-tail assets with smaller market capitalization through the automatic market maker (AMM) model.

The same is true for blockchain bridges, which are frequently hacked, and there is enough evidence to suggest that holding on to a multi-billion-dollar blockchain smart contract might not be a good idea. There are other options in the industry too, for example, Biconomy (cross-chain bridge infrastructure)'s Hyphen (cross-chain bridge application) has provided 178 million USD in transaction volume for more than 54,000 users, and only locked in their smart contracts $3 million.

What about borrowing? Of course, putting billions of dollars in an agreement that yields less than a bank account doesn't sound sane. But that's where we are now, because in times of low volatility, there's usually less appetite to borrow. If large amounts of funds provided for lending are idled, protocol yields will fall.

One metric to measure is average utilization. It is defined as the amount borrowed divided by the total capital deposited. On a platform like Aave, the average utilization over the past few years has been around 30%.

Newer native DEFI lending products like Gearbox have a capital utilization rate as high as 70%. Of course, these are two different products that cannot be compared as one is a lending product and the other provides margin, and the ability to provide yield to a larger pool of capital will start to become more important in the coming months. If you are a founder, it may be helpful to consider what changes to existing product lines could be made to reduce costs to customers or improve returns for them.When speculation around a product doesn't make sense, only two things matter (reducing costs to customers and increasing their bottom line).

The capital efficiency part becomes even more relevant when it comes to consumer-facing applications. Because you're competing with the unit economics of traditional banks and fintech companies. Unless there is an incremental drop in cost, users may not make the switch.

We've seen this happen, VCs like Moonpay and On-meta have dramatically reduced the time and cost for retail users to buy small amounts of cryptocurrencies, which used to take hours and at least 20- $30 in funding. I tried On-meta the other day and they dropped the cost to $4 and a few minutes.

first level title

industry construction

We often talk about waiting for billions of dollars in venture capital funding to be deployed, and it is no longer uncommon for startups to raise tens of millions of dollars. But even with all that money, founders rarely have insight into who their target customers are. Today, the most sought-after data set for early-stage entrepreneurs is cost to acquire customers (CAC). Few in the industry know how much it costs to acquire transacting users, and even fewer know its historical pricing.

Historically, the absence of such data has been acceptable. Because token incentives often mean that founders don’t have to go through paid channels (such as advertising) to attract users. The agreement specifies the CAC (Cost of Acquiring Users). But how do you crack that if you don't have a token plan in your immediate roadmap? This is the conundrum most (equity) well-funded venture capital firms have to contend with.

Assuming you do have CAC data, you need to know more about user roles. Earlier, you might have thought that someone using Uniswap was just another depraved farmer. But as the industry evolves, the types of users of these products will evolve in very complex forms and tools will be needed to identify them.

For example, the general assumption is that the average mobile-based gamer is a 12-year-old trying to get them to relax after a family dinner. Conversely, many paying users in the mobile gaming industry are 40-year-old women.

Nansen recognizes this need, which is why their product relies on tagging user personas based on actions in the wallet. For example, labeling a wallet as - smart money (smart wallet) or Heavy Dex Trader (heavy DEX trader) enables on-chain sleuths to gain context behind specific transactions. A handful of new companies are "scenariosizing" the user portrait business, and Layer 3, Galxe, and Rabbithole are responsible for managing the user portraits that new startups can target through tasks.

New startups often partner with these platforms to identify and target their initial users. Instead of using historical data to identify users like Nansen, platforms like Layer 3 rank users based on actions within the platform. For example, say you have a user who completes all game-related activities on the platform; when a new game's UX hits the market, it might make sense to target that specific user. Users who establish on-chain identities on the platform benefit exponentially from every recent activity, as only a few users complete all of them.

If you're using background building as a lever to untie, it might be beneficial to look at vertical niches. For example, with the NFT craze, a new class of portfolio management and NFT discovery platforms launched. More recently, gaming as an industry has been embraced by companies such as fintech. The reason for this is that each new vertical has completely different consumers.

first level title

Ease of use and discovery

image description

Limewire PRO is a file resource sharing platform. After losing the lawsuit, it was forced to close down. It took ten years for Limewire to Spotify

One way of mitigating this is through outreach programs, such as Darren Lau's Daily Ape, which quickly became a town square, with around 55,000 people subscribing to their newsletter. Likewise, Sov's Compendium curates tools for users interested in analytics and newsletters. Both of these are engines where curated promotions are generated, drawing people to other resources.

But there is a different approach, and that is through data platforms. At scale, it looks like DApp Radar or Token Terminal. Both use on-chain data to shed light on the number of users and their activity on DApps that are prominent in the ecosystem. Likewise, DeepDAO also reveals the activities that take place in DAOs (in fact, the motivation behind these products is to expand other applications or sub-products).

Notice how these tap into specific verticals? One pursues DAOs, another shares data from DApps, and the third makes it easier to view blockchain applications through the lens of traditional financial metrics. They are all data platforms, but they repackage data to help a different set of users. Over the next few years, multiple campaign planning engines displaying the same data in other formats will become the norm.

Because you can only know the frequency of transactions or the volume of transactions in nft. You need industry-specific backgrounds that come only from individuals who have already worked in that industry. Users of these products need to compare with traditional alternatives to find out whether a Web3 product is better. RWA.xyz is an example of how real-world asset lending platforms in DeFi compare to each other.

As the industry is in an era of trust deficits, we need products that provide insight into platform usage in vertical market segments. The only way to establish legitimacy is radical transparency, but not everyone needs data to make choices. End users often want a "walled garden" where they have what they need.

This is what it means for founders who build SDKs to embed functionality like blockchain bridges, lending or trading. They take a single component (such as a transactional asset), split it up and make it available as a function to multiple businesses, like MetaMask.

first level title

building moat

Let's recall what I just said, the market is oscillating between unbundling and converging because of the potential illiquidity of capital and attention in a nascent industry. During periods of peak liquidity, it is possible to start an aggregator and find attention. But when no one cares, it's better to focus on the small segment of users who need you due to lack of alternatives. These users are stickier, spend more, and provide valuable feedback that helps expand the product.

The core of splitting is to throw away the lace in the product, and it is vertical to a certain market segment. It gives businesses an edge over incumbents because it changes the unit economics of serving customers. When a venture business struggles with sales channels, getting more money from each user can be the difference between closing down or raising follow-on funding. Vertical specialization enables founders to charge more due to a lack of choice for consumers.

New entrants cannot replicate the proprietary models used to identify users (like Nansen), nor can they replicate repeat customers overnight on a platform like Layer 3. The business relationships that SDK vendors have built up over the years are hard to compete with. So, in essence, splitting makes it harder for early-stage startups to compete.

Everything I say assumes that we will struggle with a lack of speculative appetite and a lack of trust in the tools we build. A bear market is not just a drop in numbers, we die as irrelevance before we become insolvent. Segmentation is about making the product relevant enough to users in the vertical that they keep coming back to the product you're building. At the same time, and differentiate risks sufficiently to avoid competing for the same limited attention.