88% of Airdrops Fail and Nobody in Web3 Wants to Admit It

There’s a ritual in Web3 and it repeats every single cycle. A project launches a testnet and the wallet counts explode, then the team posts a “500K users!” tweet with rocket emojis. After that Venture capitalists nod approvingly, and the founders tell reporters of "unprecedented success". Then the mainnet launches and the numbers quietly decline. It is the dominant pattern in Web3 launches, and the industry has collectively decided to pretend it isn’t happening. Let’s talk about why. The Numbers That Sound Good but Mean Nothing In the first half of 2025, dozens of major testnet airdrops launched almost simultaneously, ZetaChain, Scroll, Fuel, and others, each posting wallet counts in the hundreds of thousands. ZetaChain alone accumulated over 650,000 testnet wallets . Scroll surpassed 500,000 testnet participants. Projects like Fuel reached 180,000+ participants . Now look at what happened after token launch: A Keyrock analysis of 62 airdrops across six chains found that 88% of airdropped tokens declined in price, with most experiencing the biggest drops within the first 15 days. A separate DappRadar report covering $20 billion in token distributions since 2017 found the same figure: 88% of airdropped tokens lose value within three months. Memento Research , analyzing 118 token generation events in 2025, found that 84.7% of launched tokens were trading below their TGE valuation. Nearly 64% of airdrop recipients sell immediately at the TGE. And yet, founders keeping testnet wallet counts in pitch decks. Investors keep treating them as a proxy for demand and the cycle repeats. The Mechanism: Incentive Distortion at Scale To understand why testnet numbers are unreliable, you have to understand what testnets actually incentivize. On a testnet, tokens are free. Gas fees don’t exist in any meaningful form, there is no real economic skin in the game. What does exist, very visibly, is the promise of a future airdrop. The result is a category of user that the industry calls a “testnet farmer”: someone who interacts with your protocol not because they want to, but because they’ve done the math on expected airdrop value and decided the interaction cost is worth the potential reward. The testnet farmer could not be happy with your product, they’re just checking boxes (complete task → receive points → claim tokens → sell). The entire incentive structure points away from genuine engagement and toward optimized extraction. Worse, the infrastructure for this extraction has professionalized. There are guides, communities, YouTube channels, and Telegram groups dedicated entirely to systematized testnet farming. Airdrop Alert, CryptoRank, and dozens of similar platforms publish step-by-step farming strategies for every active testnet. When a project sees 500,000 wallets engage with their testnet, a meaningful portion of those wallets belong to users who followed the same farming playbook for six other testnets that same month. The behavior developers hoped to simulate has been colonized by people optimizing for something else entirely. The Sybil Problem: When Millions of Users Are Hundreds of People The distortion of incentives is getting worse by Sybil attacks (the practice of creating multiple fake wallet addresses to increase the distribution of airdrops). LayerZero’s 2024 airdrop is the most documented case study. The protocol had 6 million unique wallet addresses interact with it heading into the token snapshot. A reasonable person might interpret this as 6 million users. But the reality is: LayerZero’s own CEO Bryan Pellegrino publicly estimated that “only 400,000 to 600,000 out of 6 million user accounts were real”. The team identified over 2 million addresses as potential Sybils initially, eventually flagging 803,093 as confirmed Sybil addresses (~13% of the total eligible pool). \ Users were scripting wallet creation, splitting capital across hundreds of addresses, and simulating legitimate transaction patterns. A separate on-chain analysis of over two million airdrop addresses found that 64% of recipients sold their tokens immediately at TGE. Another example is Optimism. They had to disqualify over 17,000 wallets for Sybil activity. ZkSync Era discovered thousands of wallets transacting closed-source tokens among themselves on a private DEX to generate the appearance of organic activity. Linea filtered roughly 800,000 wallets before its distribution. Anti-Sybil measures have become a standard component of every major airdrop, which is itself an admission that the testnet user numbers that preceded them were largely fiction. The Community That Isn’t Beyond Sybils, there is a second, harder-to-detect problem: mercenary participation. These are real humans, using real wallets, who engage with your testnet for purely extractive reasons and have zero intention of becoming actual users. The data on post-airdrop behavior is damning. ZkSync’s active address count went from over 110,000 in July 2024 to approximately 41,000 by December (85% collapse in the months following its token distribution). TVL dropped from nearly $200 million to $128 million almost immediately after the airdrop announcement triggered community backlash about allocation fairness. The network that had appeared to have enormous traction turned out to have been propped up by users who left the moment their tokens arrived. Users who earned points through mercenary activity didn’t vote on proposals, didn’t engage in governance, didn’t behave like stakeholders. They behaved like extractors, because that’s what the incentive structure had selected for. Fake Ecosystem Confidence: What VCs Are Actually Buying The downstream consequence of inflated testnet numbers is ecosystem confidence built on sand. When a project reports “500,000 testnet participants”, that number flows into pitch decks, and gets quoted in press coverage. It shows how much a project can raise, at what valuation, and on what timeline. Investors who don’t look closely at the mechanics behind the number treat it as evidence of genuine demand. This creates a compounding problem. An inflated valuation based on fake traction means the FDV at token launch is too high. A high FDV means the token needs extraordinary buy-side demand to hold its price. That demand doesn’t materialize because the mercenary users who drove testnet activity have already sold. The token collapses, the protocol struggles to retain real users, and the team spends the next year trying to rebuild with incentive programs that kick off the cycle again. Keyrock’s analysis identified high FDV as the single most damaging factor in airdrop performance: “High FDV suppresses growth and liquidity, leading to significant post-airdrop price declines”. The FDV gets set, in part, based on testnet metrics that overstated real demand. The industry has largely accepted this as normal but it shouldn’t. How to Actually Measure PMF in Web3 Product-market fit is a measurable state where users derive enough value from a product to return, refer others, and pay in some form. Here’s what real PMF signal looks like in Web3, and why it’s different from vanity metrics: After mainnet launch, track cohort retention at 30, 60, and 90 days. Industry data across 31 mobile app categories puts average Day 30 retention at around 5–8%. If you’re hitting that baseline organically, that’s a foundation. If you’re in single digits only because of active rewards programs, that’s not retention. A DAU/MAU ratio above 20% signals genuine stickiness, sounds like users making your product a daily habit. Industry benchmarks from a16z and Mixpanel put the cross-industry average around 20–25%. Below 10% is a serious warning sign. Below 5% on a blockchain protocol, without active incentives running, means you likely don’t have a product yet. For DeFi protocols, fee revenue is the single most honest signal in the entire ecosystem. It represents real users making real economic decisions and paying real costs. Marketing that grows TVL without growing fees is building a metric, not a business. The “what if we turned off incentives?” test. Any honest founder should run this thought experiment: if we removed all points, airdrops, and yield incentives tomorrow, what would our protocol usage look like in 30 days? If the honest answer is “near zero”, that is your real baseline. The Real Cost of the Illusion There’s a human cost here that rarely gets discussed. Investors who committed capital based on inflated testnet metrics lose money. Token recipients who claimed and held based on project promises see their holdings collapse within 15 days. Real builders who were working on genuine problems get drowned out by the noise of projects manufacturing traction. Developer talent gets allocated to projects that look good on paper and evaporate in practice. The broader Web3 ecosystem suffers too. That narrative isn’t entirely fair but the industry earns it every time it celebrates fake metrics as genuine signals. The ecosystem has collectively decided that the story of traction matters more than traction itself, that the appearance of PMF is more fundable than PMF, and that the gap between those two things can always be papered over with the next incentive program. The numbers that matter are never the ones you posted during testnet.