From glassnode insights by Rafael Schultze-Kraft, Kilian Heeg

Introduction

User retention is a fundamental measure of business health and a key success indicator for any industry with growth ambitions. It reflects how consistently users engage with a product or service over time.

For instance, a platform might report 100,000 weekly active users, but if 80% never return after their first interaction, its growth is unsustainable. Therefore, retention effectively encapsulates engagement, loyalty, and long-term viability, all of which directly impact profitability and sustainability.

Applying User Retention to Digital Assets

While retention is widely used to evaluate traditional products and services, its application to digital assets remains largely unexplored. Here, we introduce On-chain Retention, a novel framework for assessing user engagement with digital assets using blockchain data.

Specifically, we track on-chain address activity to measure how consistently users interact with or hold a specific asset over time. By analyzing this activity, we can determine whether holders return, remain engaged, or disengage entirely - insights that are otherwise difficult to capture in traditional financial markets.

On-chain retention offers a deeper lens into user commitment and helps investors and analysts detect fundamental behavioral patterns such as:

As digital assets gain mainstream acceptance and integrate rapidly into global business, measuring on-chain retention becomes increasingly important as it offers investors a fundamental perspective on engagement beyond price charts.

💡Bottomline:On-chain retention provides deep insights into user commitment, speculation, and market trends, making it a critical new metric for digital asset analysis.

Computing Retention for Digital Assets

To measure on-chain retention, we propose two complementary approaches: Activity Retention and Holder Retention. These methods track how users engage with a digital asset over time, either through transactions or by holding balances.

Activity Retention

Activity Retention measures how consistently addresses participate in transactions. An address is classified as active if it sends or receives a transaction within a given time window (here, we use 30-day intervals). Otherwise, it is considered inactive.

At any given point in time, we track the number of active and inactive addresses and analyze their transitions between these states. Figure 1. illustrates the possible transitions, categorizing addresses into four key cohorts:

For example, if an address was inactive in the last 30-day period but participates in a transaction during the current period, it is classified as resurrected.

Figure 1. – Transition states of Activity Retention

Holder Retention

Holder Retention shifts the focus from transaction activity to asset balances. Instead of tracking whether an address is actively sending or receiving funds, this metric analyzes whether an address continues to hold a given asset over time.

Figure 2. illustrates the possible transitions. For example, an address that held an asset in the previous 30-day window and continues to do so is classified as retained. In contrast, addresses that previously held the asset but have since reduced their balance to zero are classified as churned.

Figure 2. – Transition states of Holder Retention

Note that, unlike Activity Retention, where state transitions are binary (active or inactive), Holder Retention allows for more nuanced behaviours. Since addresses can increase, decrease, or clear their balances within the same time window, additional classifications emerge. For instance, an address that acquires an asset and subsequently sells it within the same 30-day period falls into the "New & Churned" cohort.

Why Two Distinct Retention Metrics?

Activity and Holder Retention provide complementary perspectives on user engagement, each offering unique insights into how digital assets are used and held over time.

Activity Retention is particularly valuable for evaluating assets whose value is derived from utility, where sustained transaction activity signals continued adoption and relevance.

In contrast, Holder Retention reflects a long-term investment perspective, capturing the conviction of investors who choose to hold an asset over time. This is especially relevant for assets where ownership persistence can indicate confidence in the asset’s long-term value.

Given the diversity of digital assets, both retention models are essential for a comprehensive assessment. Depending on the use case - whether evaluating active participation or gauging investor commitment - analyzing both metrics provides a more holistic understanding of an asset’s fundamentals.

💡Bottomline: Activity Retention is useful for measuring transactional behavior. Holder Retentionreflects long-term investor conviction. Combining both provides a comprehensive understanding of asset fundamentals.

A Look at the Data

Activity Retention provides key insights into the fundamental differences in how digital assets are used, as shown in Figure 3., comparing BTC and ETH. BTC exhibits high "new" and "churned" cohorts, reflecting the common practice of frequently creating and abandoning addresses for privacy. In contrast, ETH has a higher share of "retained" and "resurrected" addresses, underscoring the structural differences between UTXO and account-based blockchains.

Figure 3. – BTC and ETH Activity Retention

Figure 4 illustrates how price movements can impact Holder Retention. While market conditions naturally influence retention trends, this example highlights a distinct pattern, when a significant number of UNI holders churned amid the sharp price increase in late 2024, potentially indicating investors exiting positions.

Figure 4 – UNI Holder Retention

Beyond Existing On-Chain Metrics

The concept of on-chain retention introduces a fundamentally different perspective compared to many existing on-chain metrics, which may appear similar at first glance but fail to capture user persistence over time.

Activity Retention vs. Active Addresses

Take Active Addresses, for example. This widely used metric tracks the number of unique addresses interacting with an asset within a given timeframe. However, it does not distinguish whether the activity comes from the same users or an entirely new set.

For instance, suppose Active Addresses show 100 users in two consecutive periods. On the surface, this suggests stable engagement. But if the composition of these addresses has completely changed - meaning the original users have all churned, replaced by entirely new ones - then user turnover is 100%.

This is where Activity Retention comes in. By tracking whether the same users remain active over time, it reveals true engagement and conviction - insights that Active Addresses alone cannot provide.

Holder Retention vs. Addresses with Balance X

A similar issue arises with metrics like Addresses with Balance X, which measures the number of addresses holding a given amount of an asset. While an increasing balance cohort might suggest accumulation, it fails to indicate whether the holders within that group are the same over time or if turnover is occurring, suggesting a lack of loyalty.

In theory, the entire set of holders could be constantly rotating - meaning, what appears to be a steady increase in a holding cohort might actually be a continuous churn of funds among different addresses. Holder Retention solves this by explicitly tracking whether specific addresses continue holding over time, distinguishing between committed investors and new entrants.

How Holder Retention Expands on HODL Waves

The closest existing metrics to Holder Retention are the metrics of the HODL Waves family, which quantify the supply that has remained unmoved for a certain period. Since it inherently tracks how long assets remain in addresses, it captures elements of retention.

However, Holder Retention goes further. While HODL Waves metrics only measure unmoved supply, Holder Retention also accounts for:

By providing these additional layers of insight, Holder Retention offers a richer, more nuanced view of user behaviour beyond just hodled supply.

Retention Metrics Live in Glassnode Studio

Retention metrics are now available in Glassnode Studio, providing deeper insights into user engagement, loyalty, and market sentiment. The initial release covers BTC, ETH, and all ERC20 tokens, with more assets to follow.

Explore retention analytics today to better understand long-term user behaviour in digital assets.