The Data Problem In Web3

Web3 was supposed to solve the data problem.

Every transaction is recorded on-chain. Every wallet interaction is indexed. Every protocol movement is tracked across multiple dashboards. We have more data than ever before in the history of finance and technology.

So why are so many protocols still making decisions based on gut feelings and vanity metrics that mean almost nothing?

In our latest podcast episode, I sat down with Joel Obafemi, a DeFi analyst working with Moonwell Protocol, and Chris, a blockchain data consultant who's built dashboards for some of the top protocols in the space.

What started as a technical discussion about Dune Analytics turned into something much more interesting: an honest conversation about why having data and actually using it are two completely different things, beginning with:

The 80/20 Rule of Web3 Data

Chris dropped a truth bomb early in the conversation that set the tone for everything else: roughly 80% of protocols are still playing the vanity metrics game, while only 20% have figured out how to use data to actually drive decisions.

This isn't about intelligence or resources. It's more behavioral; most humans, and by extension, most protocols run by humans, lean toward metrics that confirm what they want to believe rather than metrics that reveal uncomfortable truths.

The result? Protocols celebrating "1 billion transactions processed!" when 99% of those are bot activity. Teams brag about total active users when 90% are one-time wallets that never return. Founders are pointing to TVL numbers that evaporate the moment you look at how that value is actually composed.

The truth is, these kinds of metrics are vanity-oriented, and it’s something we need to talk about more in the web3 industry. What these kinds of metrics are, and why they are flawed. Let's start with everyone's favorite metric:

Total Value Locked

Joel, who works deep in the risk analysis space for lending protocols, didn't hold back:

He pointed to the Ethena situation as a perfect example. On paper, Ethena had massive TVL that looked incredibly impressive.

But underneath? Much of it was generated through looping strategies, users borrowing against their deposits to deposit more, creating an inflated number that didn't represent actual unique capital at risk.

The question isn't just "how much is locked?" The real questions are:

• What percentage represents measurable liquidity when things go wrong?

• How concentrated is this value across wallets?

• What portion can actually move if needed?

• Is this organic capital or incentivized deposits that'll disappear when rewards dry up?

Surface-level TVL tells you almost nothing. But teams keep leading with it because big numbers feel good.

Cumulative Sums

Joel brought up another example that gets thrown around constantly: cumulative totals or sums.

It's a meaningful milestone, don't get me wrong. But it's not a health metric. It's a historical marker. The cumulative sum of all deposits ever made tells you about longevity and total volume over time. It doesn't tell you about current protocol health, user retention, or whether growth is accelerating or stalling in the same period of the sum.

These metrics thrive on crypto Twitter because they're easy to celebrate. But try finding them in the actual internal forums where protocols make real decisions. They're usually nowhere, because operators know these numbers don't help them do their jobs better.

Total Active Users

This is the metric that means everything and nothing at the same time. And it is my personal frustration. "Active users" or "active addresses" sounds meaningful, but it is not; it is the same as total followers on X. However, it can be meaningful if you dig deeper and start asking:

• Are these new users or returning users? Or Both

• What actions are they taking?

• How many are power users vs. one-time visitors?

• What's the distribution? (Is it 1,000 users each doing 10 transactions, or 10 users doing 1,000 transactions each?)

One whale or bot can generate address activity that looks like healthy user growth if you're not careful about how you're counting and filtering.

These kinds of insight create some of the data usage problems in the space.

What Happens When You Actually Use Data Right

Joel's perspective from the risk analysis side was illuminating because it shows what's possible when teams take data seriously.

In lending protocols specifically, the difference between good data analysis and lazy metrics is the difference between surviving a market downturn and getting liquidated into oblivion. Risk teams don't care about vanity metrics. They care about:

• Actual liquidity depth by asset

• Concentration risk across wallets

• Historical volatility patterns

• Correlated asset exposure

• Liquidation cascade scenarios

This is data that drives decisions. This is an analysis that has consequences. And it requires going well beyond what's easy to measure.

So I asked Chris what he sees as the biggest difference between teams that have dashboards and teams that actually use them effectively. His answer cut to the core:

Think about that. Dashboards without questions. Data without hypotheses.

Chris shared an example that perfectly illustrates this:

And that creates the framework of how to use data right:

1. Define a specific business problem: Not "we need data" but "we need to understand user retention by cohort."

2. Create a testable hypothesis: "Users acquired through liquidity mining contribute 40% less lifetime value than organic users."

3. Build focused analysis: Query the specific data needed to prove or disprove this

4. Make a decision based on findings: Adjust acquisition strategy, change incentive structures, etc.

Without this structure, you're just building pretty charts that nobody acts on.

What's Next

We've established the problem: most Web3 teams are measuring the wrong things, and even when they have the right data, they don't know how to use it effectively. The gap between having dashboards and having insights is wider than most people realize.

But here's the good news: this is a solvable problem. You don't need a PhD or years of experience to start doing real data analysis that drives decisions.

In Part 2 of this series, we'll show you exactly how to break out of the vanity metrics trap and learn real data analysis, even if you're starting from zero. We'll cover:

• The complete learning roadmap that works in 2026 (with AI as a tool, not a crutch)

• Why writing SQL queries by hand makes you better than relying on AI

• How to pick your niche and go deep fast

• The content multiplier effect (turning one dashboard into dozens of insights)

• Why showing your work is the hardest part (and how to do it anyway)

Continue the Conversation

Watch and Listen to the full podcast episode on YouTube.

Watch and Listen to the full podcast episode on Spotify.

If you’re new here, subscribe so you don’t miss out.