Same Volume, 46x Fewer Users: The Builder Code Retention Puzzle

Phantom has routed $44.8 billion in all-time volume through Hyperliquid with 153,128 users. Insilico has routed $36.3 billion with 3,339 users. Nearly identical volume, 46x fewer users.

That gap is not a data error. It is the clearest signal in the Hyperliquid builder ecosystem about what makes some frontends sticky and others transient. Figures throughout this article are as of July 2026, from HyperTracker's builder leaderboard.

Builder codes on Hyperliquid function like referral tags for frontends, wallets, and trading tools. Every trade routed through a builder code generates revenue for that builder, so the code becomes a proxy for product adoption: how many users chose this frontend, how much they traded, and how long they stayed. When you compare the top 15 builder codes by revenue, the patterns in user count, volume, and revenue per user reveal three distinct product archetypes that attract and retain traders in fundamentally different ways.

This piece breaks down those archetypes using real builder leaderboard data, shows why volume-per-user is a better retention signal than raw user count, and walks through how to query builder analytics programmatically.

The leaderboard at a glance

The top 15 builder codes by all-time revenue account for the vast majority of Hyperliquid's builder ecosystem. Here is how the leaders stack up across three dimensions: revenue earned, users onboarded, and volume routed.

| Builder | Revenue | Users | Volume | Vol/User | | --- | --- | --- | --- | --- | | Phantom | $23.6M | 153,128 | $44.8B | $292K | | Based | $15.2M | 42,967 | $44.9B | $1.05M | | MetaMask | $8.1M | 52,534 | $9.0B | $172K | | PVP | $8.0M | 28,223 | $17.1B | $607K | | Insilico | $3.7M | 3,339 | $36.3B | $10.9M | | Infinex | $2.8M | 9,283 | $5.2B | $555K | | Axiom | $2.4M | 34,093 | $23.0B | $676K | | TreadFi | $2.2M | 4,835 | $11.1B | $2.3M |

The volume-per-user column is where the story lives. Insilico's users each generate an average of $10.9M in lifetime volume. Phantom's users average $292K. Based sits in between at $1.05M. These are not rounding errors, they reflect completely different user bases, product types, and retention dynamics.

Three archetypes in the builder ecosystem

The volume-per-user metric sorts builders into three clean categories. Each archetype attracts a different kind of trader, retains them through different mechanisms, and generates revenue through a different model.

The gateway: high users, moderate volume per user

Phantom (153K users, $292K/user) and MetaMask (52K users, $172K/user) are wallet-first products. Users do not come to Phantom specifically to trade Hyperliquid perps. They come because Phantom is their wallet and Hyperliquid is one of many protocols they interact with through it. The builder code captures everyone who routes through the wallet, which includes casual traders who might execute a handful of trades per month alongside power users who trade daily.

The retention mechanism here is ecosystem lock-in. A user stays on Phantom because their entire DeFi life runs through it: token swaps, NFT transactions, bridging, staking, and yes, perps. Hyperliquid is a feature inside a broader product. The volume per user is lower because most of these users are not perp specialists. They dip in, trade, and move on to other activities.

This archetype generates revenue through sheer breadth. Phantom's $23.6M in builder revenue comes from having the largest user base in the ecosystem. Each individual user contributes modestly, but 153K of them add up to the top spot on the leaderboard.

The power tool: low users, extreme volume per user

Insilico (3,339 users, $10.9M/user) and TreadFi (4,835 users, $2.3M/user) represent the opposite end. These are specialized trading tools built for a narrow audience of high-frequency or algorithmic traders. The user counts are small because the products are technical, often require API integration, and are not designed for casual use.

The retention mechanism is performance dependency. An algo trader running a strategy through Insilico's infrastructure does not switch frontends because switching means re-engineering their execution pipeline. The cost of leaving is measured in development hours and strategy downtime, which creates stickiness that no UI polish can match. These users trade constantly, generating millions in lifetime volume per account, because trading is their full-time operation.

Mass (1,054 users, $2.4M/user) fits this archetype too. Barely a thousand users, but each one averages $2.4M in volume, which suggests a concentrated base of serious traders who chose the product for a specific technical capability.

The hybrid: balanced users and volume

Based (43K users, $1.05M/user), PVP (28K users, $607K/user), and Axiom (34K users, $676K/user) sit in the middle. These products attract tens of thousands of users, each generating meaningful volume. The products are accessible enough to onboard non-technical traders but feature-rich enough to retain active ones.

Based is the standout here. With $44.9B in all-time volume, it matches Phantom dollar for dollar, but with 3.6x fewer users. That means each Based user is roughly 3.6x more active than a Phantom user on average. The product has found a way to attract traders who stick around and trade frequently, which is the hardest balance to strike in this ecosystem.

PVP's position is notable too. Joined in July 2024, making it one of the oldest builder codes in the top 15, and it has accumulated $8M in revenue with 28K users. Longevity combined with sustained volume suggests a product that has survived multiple market cycles without losing its user base.

Revenue per user tells a different story than total revenue

Revenue rankings and retention rankings are not the same thing. Phantom leads total revenue at $23.6M, but its revenue per user is $154. Insilico's revenue per user is $1,113, more than 7x higher. Based comes in at $354 per user.

For builders evaluating their own product's health, revenue per user is the metric that matters for sustainability. A frontend with 1,000 users each generating $500 in builder fees is on more solid ground than one with 50,000 users generating $10 each, because the concentrated base is less likely to churn in response to a competing product launch or a fee restructuring. Those users are there because the product does something specific they need, and that functional dependency is harder to disrupt than brand awareness.

The practical question for any builder is: where do your users sit on this spectrum? If you are a gateway, your moat is distribution. If you are a power tool, your moat is technical depth. If you are a hybrid, you need both, and the risk of losing either one is what keeps you sharp.

What the join dates reveal about ecosystem maturity

The builder leaderboard includes join dates, which add a time dimension to the retention picture. PVP (July 2024) and Okto (April 2024) are the earliest entrants in the top 15. Phantom, Based, and Dreamcash joined in mid-2025. MetaMask, Infinex, and Rabby arrived in August 2025. Minara AI joined as recently as December 2025.

Two patterns emerge. First, the wallet integrations (Phantom, MetaMask, Rabby) all arrived in a cluster during mid-to-late 2025, which likely reflects Hyperliquid enabling or promoting wallet-based builder codes around that time. Second, the power tools (PVP, Insilico, Okto) were early, which makes sense because algorithmic traders and infrastructure builders are typically the first to build on a new protocol.

Revenue velocity (revenue divided by months active) shows a similar split. Phantom has been live since June 2025, roughly 13 months, and earned $23.6M, meaning about $1.8M per month. PVP has been live for 24 months and earned $8M, about $333K per month. Phantom's monthly revenue run rate is 5x higher, which reflects the power of wallet distribution even more than the raw revenue number does.

Querying builder retention data yourself

The builder leaderboard is available through HyperTracker's API. The /builders/list endpoint returns every active builder code along with their revenue, user count, volume, and join date. You can filter by timeframe (all-time, monthly, weekly) to separate permanent users from one-time visitors.

import requests

API_BASE = "https://ht-api.coinmarketman.com/api/external"
headers = {"Authorization": "Bearer YOUR_JWT_TOKEN"}

# Get all-time builder leaderboard
builders = requests.get(
    f"{API_BASE}/builders/list/timeframe/all",
    headers=headers
).json()

# Calculate volume per user for each builder
for b in builders[:15]:
    vol_per_user = b["volume_usd"] / max(b["users"], 1)
    rev_per_user = b["revenue_usd"] / max(b["users"], 1)
    print(f"{b['refCode']:15s}  users={b['users']:>8,}  "
          f"vol/user=${vol_per_user:>12,.0f}  "
          f"rev/user=${rev_per_user:>8,.0f}")

For retention analysis specifically, compare the all-time endpoint against the monthly or weekly timeframe. A builder code with 50,000 all-time users but only 2,000 monthly active users has a very different retention profile than one with 5,000 all-time users and 4,000 monthly active. The ratio of monthly to all-time users is a rough proxy for stickiness, and tracking it over time shows whether a frontend is growing its active base or living off historical signups.

You can also cross-reference builder data with cohort analytics. Pull the wallets associated with a builder code, then check which behavioral cohorts those wallets belong to. A frontend whose users are concentrated in the Money Printer and Smart Money cohorts is retaining high-value traders. One whose users cluster in Exit Liquidity and Semi-Rekt is retaining unprofitable traders, which is worse than not retaining them at all from an ecosystem health perspective.

Implications for builders and analysts

If you are building a trading frontend on Hyperliquid, the builder leaderboard tells you where the ecosystem is crowded and where it is not. The gateway archetype is dominated by established wallets with enormous distribution advantages. Competing with Phantom on user count is impractical for a new entrant. The opportunity is in the power tool and hybrid spaces, where product differentiation matters more than existing install base.

For analysts and fund managers, builder code data is a leading indicator for ecosystem health. A protocol where most volume flows through a few concentrated builder codes is more fragile than one with distributed volume. If the top 3 builder codes account for half of all volume, a single integration dropping could meaningfully impact the protocol's liquidity. Tracking builder code concentration over time is a useful complement to standard DEX volume metrics.

For traders, knowing which frontend you are using and how its user base behaves gives you context on execution quality. A power tool frontend with 3,000 users is routing you alongside serious algorithmic traders. A gateway with 150,000 users is routing you alongside everyone from professional market makers to first-time DeFi users opening a small position. The execution environment differs, and it can affect fill quality during volatile periods.

Track Builder Code Metrics Programmatically

HyperTracker's API gives you the full builder leaderboard with revenue, users, volume, and historical timeframes. Query builder codes, cross-reference with wallet cohorts, and build retention dashboards on our data.

Explore the Free Tier

Retention on Hyperliquid is not about having the most users. It is about having the right users for your product type, and keeping them active. Phantom wins on breadth. Insilico wins on depth. Based wins on balance. The builder code data makes all of this visible, and the frontends that learn to read it will build better products than the ones that only count signups.