Most Web3 loyalty programs are built on the same idea that keeps failing in Web2 loyalty: reward activity and call it loyalty. Complete a quest, claim a token, repeat. The problem isn't the blockchain. It's that activity and loyalty are different things, and the programs that confuse them keep closing.
Starbucks shut its Odyssey NFT loyalty programme by the end of March 2024, less than a year and a half after its beta launch, citing complexity that never matched the simplicity of the core Starbucks Rewards programme members already used. Nike wound down RTFKT, its flagship Web3 studio, in January 2025 and quietly sold it off in December 2025 as new leadership pursued a "back to basics" strategy. Both companies still run enormously successful loyalty programmes. Starbucks Rewards alone counted 33.8 million active US 90-day members in its most recent quarterly filing at the time, up 4% year over year, and has historically driven around 59% of company-store spend. The core programmes didn't need quests or collectibles to work. They needed a real reason for the customer to come back, and they already had one.
The pattern: quest farming produces sign-ups, not loyalty
Quest farming is the practice of completing the minimum tasks required to qualify for a token or reward, with no intention of engaging with the product afterwards. It's the Web3 evolution of a problem loyalty marketers have fought for decades: rewarding a behaviour that's easy to fake.
The data on what happens once the reward is claimed is consistent and severe. When Dune Analytics studied Uniswap's landmark UNI airdrop, of roughly 220,000 eligible wallets, only 6.7% still held any UNI a year later. More than 75% had sold within a week, 80% within a month, and around 98% never once participated in governance, the entire stated purpose of the token. Other 2020–21 airdrops told the same story: 1inch retention sat at 7.9%, and LooksRare saw roughly 85% of recipients dump despite built-in staking incentives designed to discourage exactly that.
It isn't universal. The same Dune analysis found ENS holders retained at 23.9% and Hop Protocol at 38.7%, both markedly higher, and both tied to airdrops that leaned more on real product usage than pure task completion. The gap between "claim and leave" and "claim and stay" tracks closely with how much the reward depended on something the user actually had to use the product for.
Sybil farming makes the picture worse. LayerZero identified enough suspected multi-accounting ahead of its 2024 airdrop that it gave farmers a self-report deadline; those who complied kept only 15% of their allocation. Daily cross-chain transaction volume on the network fell by more than half in the weeks after the snapshot, evidence that a large share of "activity" evaporates the moment the incentive does. Arbitrum's team later identified nearly 149,000 confirmed sybil addresses that had passed its eligibility checks, capturing an estimated 21.8% of the airdrop.
The controlled evidence: rewarding real usage changes the outcome
The clearest test of the theory came from Optimism's Airdrop 5 in October 2024, and it's worth sitting with because it's a genuine natural experiment, not a correlation. Optimism distributed 10.4 million OP tokens to 54,700 addresses and later analysed the results with researchers including Andy Hall of Stanford Graduate School of Business. Simply receiving the airdrop raised 30-day retention by 4.2 percentage points. But the bonus categories split cleanly by what they actually rewarded: bonuses tied to genuine cross-chain product usage raised retention further, by as much as 10 percentage points for the strongest category. A bonus that rewarded frequent activity, closer to farming behaviour, cut 30-day retention by 7.1 percentage points. The researchers' own explanation: farmers who are rewarded for activity sell their tokens and leave; users rewarded for real usage stay.
That's the whole argument in one study. Points measure attention. Verified action measures whether something actually happened.
Web2 already has this problem too
This isn't a crypto-specific failure mode. The Bond Loyalty Report, the industry's longest-running study (its 2025 edition drew on 37 million data points from over 250,000 consumers), found the average consumer is enrolled in 17.4 loyalty programmes but actively uses only 8.8 of them. McKinsey's loyalty research puts more than half of all programme memberships in a dormant state, while active members spend around 10% more than inactive ones and members who actually redeem rewards spend around 25% more. Bond's 2026 report names the mechanism directly: a widening gap between "artificial loyalty, driven by systems, signals, and scale" and "emotional loyalty, built through genuine human connection," warning that too often, "activity is mistaken for affinity."
Web3 didn't invent the gap between activity and loyalty. It just made the gap easier to measure, because every claim and every dump happens on a public ledger.
Points measure attention. Zenko measures action.
A points system or a quest platform counts what a user did inside an app: clicks, quiz answers, social follows, logins. None of that requires the reward to mean anything to the person doing it. Zenko's model works differently: the token and the real-world outcome behind it are only released once a verified action has genuinely happened. That might be a completed course, a purchase, a distance walked and confirmed, or attendance at an event. HP saw a 35% reduction in cost-per-lead running a reward budget worth just 5% of campaign spend; Lenovo moved cost-per-lead from $140 to $90 with a $10 reward tied to a completed action rather than a click. Neither number depends on someone farming a task list. Both depend on someone actually doing the thing.
Verified action isn't free of trade-offs. It adds friction the pure-quest model doesn't have, and building the verification layer is genuinely harder than issuing points. Some quest platforms serve a real purpose, too: Galxe and Zealy are useful for structured discovery and onboarding, not every task-based mechanic is farming. The distinction that matters is whether the reward exists to prove attention happened, or to prove action did.
A five-question check for any loyalty programme, Web3 or not
Before adding token or point mechanics to a loyalty programme, or evaluating whether an existing one is working:
- Would the user still do this action if there were no reward attached?
- Does completing the "task" require anything real from the user, or just a click?
- Is there a verifiable outcome behind the reward, or only a balance going up?
- What happens to engagement in the week after the reward is claimed?
- Could this be farmed by someone with no intention of ever using the product again?
If the honest answer to the last question is yes, the programme is measuring attention, not loyalty.
Where this series goes next
This piece sets out the pattern. The five pieces alongside it go deeper on each part of the picture:
- What Is a Web3 Loyalty Program? — a practical definition and a closer look at why the highest-profile examples failed.
- Blockchain Loyalty Programs: The Real Benefit Isn't Interoperability — most vendor content sells the wrong benefit; sybil resistance is the one that actually protects a loyalty budget.
- Crypto Loyalty Programs vs Reward Tokens — why STEPN's collapse is a warning for any "earn" model, including ones built on real action.
- What Is Quest Farming? — a full definition, the sybil data, and what it costs a brand.
- Verified Real-World Actions — the model itself, what it borrows from proof-of-attendance and proof-of-personhood work, and what it looks like in production.
FAQ
Are all Web3 loyalty programs quest farming? No. Programmes tied to verified product usage, real purchases, or confirmed real-world actions behave differently from pure task-completion or points systems. The Optimism Airdrop 5 data shows the two produce opposite retention effects.
Is quest farming a Web3-only problem? No. It's a sharper version of the "activity ≠ loyalty" problem Web2 loyalty programmes have had for years, evidenced by McKinsey and Bond's research on dormant memberships. Web3 just makes it visible on-chain.
What's the single best piece of evidence that verified action beats points? Optimism's Airdrop 5 analysis (January 2025), because it's a controlled comparison within the same programme: rewarding real cross-chain usage raised 30-day retention by up to 10 percentage points, while rewarding frequent activity cut it by 7.1 points.