content
Key Takeaways
Every retention intervention should be measured: track reactivation within 14 days for win-back campaigns and retention rates for alert-triggered outreach.
The goal is a closed feedback loop: signal detected, intervention triggered, outcome measured, threshold or message refined.
DeFi's retention problem is structural but solvable. The apps that retain users are watching earlier signals.
Five onchain signals precede churn: declining transaction frequency, full position exits, withdrawal velocity spikes, post-failure inactivity, and cross-protocol migration.
Declining transaction frequency is the earliest and most consistent churn signal. A wallet whose 7-day transaction count falls to 30% or below its 30-day average warrants a flag.
TVL dashboards show the past. Wallet-level behaviour data shows the next 14 days.
TVL is a weak churn signal because a flat reading can mask whale concentration, price appreciation effects, or a broad exit by smaller active wallets.
Churn in DeFi is behavioural, not event-based. There is no cancellation event. A wallet churns when it stops returning within the app's expected activity window.
Churn thresholds vary by app type: a DEX typically uses 14 or more days without a swap; a lending app uses loan repayment with no new position opened within 30 days.
Prioritise at-risk wallets by lifecycle stage and position size. Wallets in the At Risk stage that sit in the top 20% by deposited value warrant immediate attention.
New wallets showing zero deposit activity after their first session and resurrected wallets showing repeat disengagement are the next priority tiers after high-value at-risk wallets.
Formo's At Risk lifecycle stage automates wallet flagging with configurable inactivity thresholds per project via Lifecycle Settings.
DeFi apps typically discover churn through lagging indicators: TVL drops, active wallet counts fall, and the dashboard confirms a trend that is already weeks old. By the time aggregate numbers move, the users behind them have long since left.
Metrics like TVL aggregates deposited value across all wallets into a single number. An app can show flat TVL while losing its most engaged users, masked by price appreciation or a handful of whale positions. Wallet behaviour is the leading signal. TVL is the confirmation.
The retention data is severe. Only 2–4% of DeFi users remain active after one year, with most apps losing the majority of new wallets within 90 days.
This guide covers:
What an at-risk wallet is and how churn works in DeFi
Why TVL fails as a churn signal and which metrics to use instead
5 behavioural signals that consistently precede churn, with sourced measurement guidance for each
A segmentation framework for prioritising which at-risk wallets to act on, drawn from Formo's lifecycle definitions
A 5-step workflow for turning signals into interventions
Key takeaway: Churn in DeFi is predictable. At-risk wallets leave a trail of behavioural signals days or weeks before they exit. The teams that retain users are the ones who read those signals early enough to act.
What Is an At-Risk Wallet in DeFi?
In DeFi, the wallet is the unified account of the user. Every interaction, deposit, swap, borrow, or withdrawal is a signed transaction from a wallet address. When a wallet stops interacting with your app, that is churn.
Churn in DeFi means a previously active wallet has stopped engaging with your app within a defined time window. Unlike subscription products, where churn is a cancellation event, DeFi churn is behavioural: the wallet simply goes quiet. There is no cancellation, no notification, and no record of intent.
An at-risk wallet is one that was previously active but is showing early signs of disengagement before it exits entirely. This includes wallets with declining transaction frequency, wallets that have partially withdrawn a position, or wallets that encountered a friction event and did not return. The defining characteristic is that the wallet has not yet churned, but its behaviour suggests it is heading that way.
The distinction matters operationally. A churned wallet has already left. An at-risk wallet can still be retained, but only if the signal is caught early enough to act on.
According to Blockonomi's 2026 market analysis, only 2–4% of DeFi users remain active after one year. The gap between a retained wallet and a churned one most often comes down to what happened in the first 7 days, and whether anyone was watching.
Why TVL Fails as a DeFi Churn Signal
TVL remains the most-cited metric in DeFi, but it is structurally ill-suited for predicting user churn. It aggregates deposited value across all wallets into a single number. When that number stays flat, it could reflect a stable user base, a handful of whale wallets offsetting an exodus of smaller active users, or token price appreciation masking net outflows at the wallet level. Without wallet-level data, those scenarios are indistinguishable.
A 2026 systemic risk analysis published on arXiv found that TVL-based metrics consistently lag behind structural fragility signals. The study's Correlation Fragility Indicator detected elevated ecosystem risk weeks before TVL movements confirmed it.
More reasons why TVL is a poor churn signal:
Concentration masking. A small number of high-value wallets can hold TVL stable while the broader user base exists. When those wallets leave, the drop is sudden and severe.
Price appreciation masking. Rising token prices inflate TVL even as wallet count and transaction frequency decline. The metric looks healthy while engagement deteriorates.
Incentive-driven distortion. Liquidity mining programs attract users without creating loyalty. TVL spikes during incentive periods and collapses when rewards end.
The apps that catch churn early track wallet-level signals: transaction frequency, position duration, withdrawal velocity, and cross-protocol activity. These move before TVL does.
The Early Warning Signals of Wallet Churn
Across DeFi apps, at-risk wallets consistently exhibit behavioural changes in the days and weeks before they exit. These signals are readable with wallet-level tracking in place.
Signal 1: Declining Visit and Transaction Frequency
The most reliable leading indicator of churn is a drop in how often a wallet interacts with your app. A wallet executing 2–3 transactions per week that drops to 1 per fortnight is showing disengagement before it exits entirely.
What to watch: Track each wallet's rolling 7-day and 30-day transaction count using Formo's Transaction Frequency chart, which shows the distribution of transaction counts per wallet. A wallet whose 7-day count falls to 30% or below its 30-day average warrants a flag.
A zScore behavioural scoring study on arXiv, which analysed Uniswap v3 LP and swap activity, found that high-retention wallets share 2 consistent traits: long-term liquidity provision and consistent transaction frequency. Frequency consistency across deposit, withdrawal, and swap events was among the strongest predictors of sustained engagement.
Signal 2: Position Exit With No Subsequent Deposit
When a wallet closes its entire position either fully or partially and opens no new one within your app's typical transaction cadence, that is an exit signal. Routine rebalancers redeposit. Churning wallets go quiet.
What to watch: In Formo's behavioural segments, build a segment for wallets that performed a withdrawal event at least once in the last 14 days AND have zero deposit events in the same window. Pair this with a net worth or position-size filter to focus on wallets above your app's median deposit.
The window should match your app's natural cadence. A DEX with daily activity warrants a tighter window than a quarterly yield vault.
Signal 3: Withdrawal Velocity Spikes
A sudden increase in withdrawal events at the wallet count level, even when TVL appears stable, is one of the most actionable pre-churn signals available. Withdrawal clustering, where multiple wallets churn within a short time window, often indicates a coordinated response to an external event: a competing pool launching higher yields, an incentive program ending, or a perceived app risk.
What to watch: Track daily withdrawal event counts by wallet count. Measuring by volume distorts the signal when whale wallets are involved. Wallet-count velocity tells you whether the breadth of your user base is contracting.
A sustained increase in the number of wallets initiating withdrawals over a 48-hour window warrants investigation, regardless of what the TVL chart shows. Formo's Activity feed lets you filter by event type and view timeseries breakdowns by wallet count, making this pattern visible without custom SQL.
Signal 4: Cross-Protocol Migration
A wallet that begins interacting with a competing app while reducing activity on yours is already in the process of leaving. Cross-protocol behaviour is visible onchain and provides the clearest signal of intent.
What to watch: Each wallet profile in Formo includes an Apps tab showing the user's active DeFi positions across all major chains, including the app name, chain, USD value, and portfolio percentage for each position. If a wallet that previously held a position in your app now shows a growing position in a competing app on the same asset class, that is a concrete migration signal. Use this alongside the Audience Insights page, which shows the top apps your user base is active in, to identify whether a competing app is systematically drawing from your user base.
The Churn Signals Summary
Signal | How to measure in Formo | Threshold |
Transaction frequency | Transaction Frequency chart: rolling 7-day vs 30-day wallet count | 7-day count at 30% or below 30-day average |
Full position exit | Behavioural segment: withdrawal event ≥1, deposit event = 0, same 14-day window | 65%+ pool value removed by a single address (Chainalysis 2025) |
Withdrawal velocity | Activity page: withdrawal event timeseries by wallet count | Sustained wallet-count increase over a 48-hour window |
Cross-protocol migration | Wallet profile, Apps tab; Audience Insights top apps | New material position in a competing protocol on the same asset class |
How to Segment Your At-Risk Users by Lifecycle
Segmenting at-risk users by impact and likelihood lets your team prioritise where to spend time and resources.
Here is a user segmentation framework you can use. Four segments (At-Risk Power Users, New Users, Resurrected Users, Incentive-Driven Users) map directly to user lifecycle. Use Formo’s User Segmentation feature to define and find these users in your DeFi app.
Segment 1: At-Risk Power Users
Formo's At-Risk stage flags wallets last seen at least 14 days ago, with fewer than 5 active days in the last 30 days, and at least 1 active day in the prior 30–60 day window. These are previously active wallets whose activity has slowed but who have not yet churned.
Filter by net worth and behaviour to surface wallets where a single exit would move your TVL. Prioritize power users in this segment because they have the most impact on your bottom line.
Segment 2: New Users
Wallets first seen within the last 30 days sit in Formo's New lifecycle stage. The first 7 days are the highest-leverage window for retention. Use Formo's first-seen behaviour filters to ask questions like "users who deposited within 24 hours of first session" and identify which onboarding paths produce wallets that return.
Segment 3: Resurrected Users
Wallets that were making transactions regularly, went dormant for 30 or more days, and have recently re-engaged sit in Formo's Resurrected stage. Users who have churned before have a higher-than-average likelihood of churning again.
Study what brought them back via their attribution data in the wallet profile, so you know which re-engagement channels are working.
Segment 4: Incentive-Driven Users
Wallets that arrived during a liquidity mining or points program and have shown no engagement outside of reward collection. Build this segment in Formo by combining referrer and UTM data (campaign = liquidity program name), wallet labels, and behavioural filters. These wallets are likely to exit when incentives end.
How to Build Segments in Practice
Effective segmentation requires combining 2 data layers:
App event data. Deposit, withdrawal, swap, and borrow events from your smart contracts, aggregated at the wallet level. Formo captures these automatically via contract event tracking and surfaces them in each wallet's activity timeline.
Wallet intelligence. Cross-protocol activity, wallet age, token holdings, and net worth. Each wallet profile in Formo shows the user's full DeFi portfolio across chains in the Apps tab, alongside their onchain attestations and social profiles.
The combination makes segmentation actionable. A wallet showing declining transaction frequency within your app looks very different depending on whether its cross-protocol activity is increasing (migrating to a competitor) or decreasing (reducing overall DeFi exposure). The intervention for each scenario differs.
For teams building this from scratch, Formo's onchain user segmentation guide covers the practical implementation steps for wallet-level cohort analysis.
Formo computes lifecycle stages automatically for every wallet in your project. Lifecycle thresholds are configurable per project, so you can tune the inactivity window to match your app's cadence.
The Key Metrics That Predict Churn in DeFi
Beyond the signals covered above, there are specific quantitative metrics that DeFi teams should track weekly. Each one moves before TVL does.
Wallet Concentration Ratio
This measures what share of your app's total liquidity is held by your top wallets. A healthy app has distributed liquidity across many wallets. A fragile one has 80% of its TVL concentrated in 3 or 4 addresses.
Track this as a rolling metric. If your top-10 LP concentration is increasing week over week, you are accumulating fragility even if TVL appears stable. The OECD's 2024 analysis of DeFi liquidity concentration found that large withdrawal events by concentrated liquidity providers can destabilise entire pools, making this a structural metric worth tracking weekly, not just when TVL moves.
Wallet-Level Retention Cohorts
Group wallets by the week they first deposited or completed a transaction, then track what share of each cohort is still active at 7, 30, and 90 days. Formo's Retention chart builds this cohort matrix automatically. Select any entry and retention events, such as a transaction or a custom deposit event, to define what "retained" means for your app.
Retention benchmarks differ by app type. The following are drawn from Formo's retention docs:
App type | Week 1 | Week 4 | Week 8 |
DEX | 30–40% | 15–25% | 10–20% |
Lending | 25–35% | 15–20% | 10–15% |
If a cohort's week-1 retention sits at the bottom of these ranges, the issue is onboarding and early activation. Fixing the first-week experience produces more impact than any win-back campaign.
For a detailed breakdown of the activation metrics that drive early retention, see Formo's DeFi activation metrics guide.
Withdrawal Velocity
Track the rate of new LP deposits versus withdrawals on a daily and weekly basis, normalised to exclude outlier whale movements. A sustained withdrawal trend at the wallet count level, even when TVL appears flat, is an early warning signal.
Measure withdrawal velocity by wallet count. Wallet-count velocity tells you whether the breadth of your user base is contracting.
Revenue Per Wallet Trend
A wallet's fee contribution (through swap fees, spreads, or interest payments) is a proxy for how deeply it is using the app. A declining revenue trend across a cohort, even before transaction frequency drops, signals that wallets are using the app less intensively.
Track this at the cohort level: group wallets by acquisition month and watch whether each cohort's average fee contribution is growing, flat, or declining over time.
The Metrics Monitoring Stack
Metric | Frequency | What It Signals |
LP concentration ratio | Weekly | Structural fragility risk |
Wallet retention by cohort | Weekly | Onboarding quality and product stickiness |
Withdrawal velocity (wallet count) | Daily | Early exit behaviour |
Revenue per wallet trend | Weekly | Depth of engagement before frequency drops |
Post-failure inactivity rate | Daily | UX friction causing disengagement |
Cross-app migration rate | Weekly | Competitive pressure on your user base |
What to Do When You Identify an At-Risk Wallet
Identifying at-risk wallets is only useful if it triggers an action. The intervention needs to match the churn signal, the wallet's value, and the likely reason for disengagement.
Match the Intervention to the Signal
Declining transaction frequency (no withdrawal yet)
The wallet is still present but less engaged. This is the best window for intervention. Surface a reason to return: a new yield opportunity, a feature update relevant to their historical behaviour, or an app improvement that addresses a friction point they may have experienced.
Goal: Re-engage before the wallet reduces exposure.
Partial withdrawal detected
The wallet is actively reducing exposure. Acknowledging the rate differential directly and surfacing other positions or strategies within your app is more credible than a generic re-engagement message.
Goal: Retain and prevent full exit.
Post-failure inactivity
The wallet encountered friction. A simplified entry point or a direct support prompt is more effective than a yield-focused message.
Goal: Remove the friction that caused the drop-off.
Cross-protocol migration detected
The wallet has already moved elsewhere. Win-back messaging needs a concrete reason to switch back: a better rate on the same asset, a new feature, or a time-limited incentive.
Goal: Give a specific reason to return, not a generic ask.
Set Up Real-Time Alerts for High-Value At-Risk Wallets
For wallets in your top 20% by position size, manual monitoring is too slow. Formo's Alerts fire in real time when a wallet triggers a configured event condition, with notifications delivered via webhook or Slack. Add conditions to filter on event properties, for example, chain_id = 8453 or status = failed, so alerts only fire for the events that matter.
For example, create a large withdrawal alert. Configure a custom withdraw event with a volume threshold condition that fires when a wallet withdraws above your defined threshold.
Each alert payload includes the wallet address, which links directly to their wallet profile in Formo, giving your team immediate access to the wallet's lifecycle stage, net worth, acquisition source, and full activity timeline.
Not All Churn Is Fixable
Some wallets should churn. Airdrop farmers, Sybil wallets, and incentive extractors with no genuine app usage represent no loyalty. Retaining them costs resources and distorts your retention metrics.
The goal is to retain the wallets whose behaviour indicates genuine product engagement. The segmentation framework above helps make this distinction: focus intervention resources on wallets that showed real usage before the churn signals appeared.
For a comprehensive look at the different types of churn and which ones are worth addressing, Formo's guide to reducing customer churn in Web3 covers the root causes and retention strategies in detail.
How to Build an Automated Churn Detection Workflow
The signals and metrics above are only actionable as part of a systematic workflow. The workflow below is designed to be implementable without a dedicated data team. It assumes you have basic event tracking in place for your app's core actions.
Step 1: Define Your Churn Event
Define what churn means for your specific app before tracking anything else. For a DEX, churn is likely 14 or more days without a swap from a previously active wallet. For a lending app, it is a wallet that has repaid its loan and opened no new position within 30 days. For a liquidity vault, it is a full withdrawal with no redeposit within the vault's typical rebalancing window.
Document the definition and use it consistently. If your team measures churn differently across dashboards, cohort comparisons break down.
For a full walkthrough of how to define and measure churn events for different app types, see Formo's DeFi churn analytics guide.
Step 2: Build Retention Cohorts
Group wallets by the week they first interacted with your app. Track what share of each cohort is still active at 7, 30, and 90 days. Run this analysis weekly. If a cohort acquired after a specific product update shows materially better 30-day retention than previous cohorts, that is evidence that the change worked.
Step 4: Automate At-Risk Flagging
Once your wallet-level data is in place, automate the flagging of at-risk wallets based on the thresholds from the signals section. The goal is a daily or weekly list of wallets that have crossed a churn risk threshold, ranked by position size or app revenue contribution. This list should feed directly into your retention intervention workflow.
Step 5: Close the Loop With Outcome Tracking
Every intervention should be tracked. When you send a win-back campaign to dormant wallets, measure what share is reactivated within 14 days. When you set up a real-time alert for high-value wallets, measure whether the alert-triggered outreach retained the wallet. The goal is a feedback loop: signal detected, intervention triggered, outcome measured, threshold or message refined.
Formo's At Risk lifecycle stage automates this flagging. Thresholds are configurable per project via Lifecycle Settings.
The full DeFi retention guide on Formo's blog covers the measurement framework for tracking whether retention efforts are producing real results.
Summary
DeFi's retention problem is structural, but it is solvable. The apps that retain users are watching earlier.
The 5 signals covered in this guide, declining transaction frequency, full position exits, withdrawal velocity spikes, post-failure inactivity, and cross-protocol migration, are all visible in your onchain data before a wallet churns. The segmentation framework gives you a way to prioritise which signals to act on first. The workflow gives you a repeatable process to turn signals into interventions.
The shift is from measuring what already happened to monitoring what is about to happen. TVL dashboards show you the past. Wallet-level behaviour data shows you the next 14 days.
For DeFi teams looking to build this kind of visibility without standing up a custom data stack, Formo's retention analytics provides wallet-level cohort analysis, real-time event alerts, and cross-protocol behavioural context out of the box.
Monitor At-Risk Wallets in DeFi with Formo
Get contract event ingestion, wallet-level lifecycle tracking, cohort retention analysis, and behavioural segmentation, without requiring a custom data pipeline.
What you get with Formo:
Automatically flag wallets in the At Risk lifecycle stage, with configurable inactivity thresholds per project via Lifecycle Settings
View each wallet's full DeFi context in wallet profiles: lifecycle stage, net worth, transaction frequency, active positions across chains, and acquisition source
Build and save at-risk cohorts using behavioural segments with AND/NOT logic across event sequences, wallet properties, and lifecycle stages, then export to CSV for outreach on socials such as X, Farcaster, or Lens
Track week-by-week cohort retention in the Retention chart
Receive real-time webhook or Slack notifications via Alerts when a wallet triggers a failed transaction or large withdrawal event
Query your retention data in plain language using Ask AI: "Which cohorts have the lowest 30-day retention?" returns a chart without writing SQL
Frequently Asked Questions
What is an at-risk wallet in DeFi?
An at-risk wallet is one that was previously active but is showing early signs of disengagement before it exits entirely. This includes wallets with declining transaction frequency, wallets that have partially withdrawn a position, or wallets that encountered a friction event. The wallet has not yet churned, but its behaviour suggests it is heading that way.
How is churn defined for a DeFi app?
Churn in DeFi is behavioural. A wallet churns when it stops returning within your app's expected activity window, with no new deposit, swap, borrow, or position update. There is no cancellation event. The definition varies by app type: a DEX typically uses 14 or more days without a swap, a lending app uses loan repayment with no new position opened within 30 days.
Why is TVL a weak signal for detecting churn?
TVL aggregates deposited value into a single number. A flat TVL reading can mask whale concentration, price appreciation effects, or a broad exit by smaller active wallets. Wallet-level signals such as transaction frequency, withdrawal velocity, and position duration move before TVL does, giving teams an earlier window to act.
Which onchain signals most reliably precede churn?
Declining transaction frequency is the earliest and most consistent signal. A wallet whose 7-day transaction count falls to 30% or below its 30-day average warrants a flag. Full position exists with no subsequent deposit, sustained withdrawal velocity increases measured by wallet count, and cross-app migration are the other 3 signals to watch.
How should teams prioritise which at-risk wallets to act on first?
Prioritise by lifecycle stage and position size. Wallets in Formo's At Risk stage that also sit in your top 20% by deposited value warrant immediate attention. New wallets showing zero deposit activity after their first session and resurrected wallets showing repeat disengagement patterns are the next priority tiers.
