An airdrop or task campaign distributes tokens or rewards in exchange for defined actions, and its only honest measure of success is how many participants are still present a month later. Most campaigns are designed backwards: the budget is set, the tasks are chosen to be easy to complete, and retention is discovered afterwards. This guide covers task design, verification, sybil resistance, and the metrics that separate a community from a queue.
Key Takeaways
- Completion rate is not a success metric. A campaign where everyone completes every task has tasks that are too cheap to filter anyone.
- Verification before counting is the whole game. A participant who is counted on click rather than on completion is not a participant.
- Measure retention 30 days after the campaign closes, not on the final day. The gap between those two numbers is the campaign’s real result.
- Sybil resistance costs either money or friction. Projects that pay for neither get the participant count they asked for and none of the community.
- The proportional lift is largest on the smallest base, which is the opposite of how most projects allocate budget.
What is a task campaign, and how does it differ from an airdrop?
A task campaign rewards defined actions such as joining a channel, holding a position, or completing an on-chain interaction. An airdrop distributes tokens to addresses that already meet a condition, often retroactively. The practical difference is timing: a task campaign shapes behavior while it runs, an airdrop rewards behavior that already happened.
Both fail the same way. The reward attracts people whose only interest is the reward, and when it stops, so do they. The design question is therefore not how to attract participants but how to make the reward correlate with something you want to keep.
How do you design tasks that filter?
Good tasks cost the participant something that a farm cannot cheaply supply: time, an on-chain action with a gas cost, or a public association with the project. Tasks that cost only a click select for the population that clicks for a living.
A workable ladder runs from cheap to costly, and the campaign learns from where people drop out.
| Task type | Participant cost | Farm resistance | What it selects for |
|---|---|---|---|
| Join a channel | None | Very low | Volume, nothing else |
| Post or reply publicly | Reputation | Moderate | Willingness to be seen |
| On-chain interaction | Gas | Moderate to high | Wallet ownership |
| Hold for a period | Opportunity cost | High | Time preference |
| Verified identity tier | Privacy and friction | Highest | Distinct people |
The identity tier is the strongest filter and the most expensive in drop-off. In community campaigns we run a KYC-verified tier where a project needs assurance that participants are distinct people, and a standard tier where completion verification is enough. Which one is right depends on whether the reward is large enough to make farming worth the effort.
Why does verification have to happen before counting?
Counting on click rather than on completion is the most common way a campaign report becomes fiction. A participant who opened a link is not a participant who did the task, and the difference between those two numbers is often several multiples.
The rule we apply is that a participant counts only after the action is verified, not when the action is started. This produces smaller headline numbers than click-counting and far more accurate ones, and it makes the 30-day retention figure meaningful rather than decorative.
Ask any supplier one question before signing: at what moment does a participant enter the count. Suppliers who count on click will tell you so, and suppliers who cannot answer are counting on click.
What does a real campaign result look like?
Three 2026 campaigns give a sense of realistic movement. Fourmeme grew from 35,000 to 47,000 members, Atleta Network from 130,000 to 141,000, and Tothemoon from 40,000 to 45,000.
Read those as ratios rather than totals. The 35,000-member community gained roughly a third; the 130,000-member community gained under a tenth. Larger bases absorb the same campaign with less proportional movement, because the incremental participant is harder to find and the existing base dilutes the effect. Projects routinely budget as though the reverse were true.
Nothing in those numbers says anything about retention, which is why retention is measured separately and later. A campaign that adds 12,000 members and keeps 3,000 has produced a different outcome from one that adds 5,000 and keeps 4,000, and the headline figure hides which one you bought.
How do you resist sybil farming without killing the campaign?
Sybil resistance is a trade between cost and friction, and there is no configuration that avoids both. Every mechanism either spends money on verification or spends participants on drop-off.
Four measures carry most of the weight. Require an on-chain action rather than a social one, because gas is a real cost per identity. Reward duration rather than the moment of joining, since farms optimize for immediate payout. Cap the reward per participant so that industrialized entry is not worth the setup. And verify identity on the tier where the reward is large enough to justify it.
What does not work is detecting farms after the fact. By the time the pattern is visible in the data, the tokens are distributed and the participant count is already in a report.
Which metrics matter?
Thirty-day retention is the share of campaign participants still present a month after the campaign closes. This is the number the campaign should be judged on, and measuring it on the final day instead makes every campaign look successful.
Completion drop-off by task shows where the ladder is filtering. A task that almost everyone completes is not filtering; a task that almost nobody completes is priced wrong.
Post-campaign participation is the share of participants who do anything at all afterwards: post, hold, or interact. A figure near zero means the campaign bought a queue, not a community.
Cost per retained participant, rather than cost per participant, is the only figure worth comparing between campaigns. The two can differ by an order of magnitude for the same spend.
What are the common failure modes?
Campaigns fail in four recognizable ways, and each is a design decision rather than bad luck.
Setting the participant target before the task design means the tasks get easier until the target is reachable. Rewarding the join rather than the behavior after the join means the reward is collected and the relationship ends. Running the campaign without a content rhythm to arrive into means participants land in a silent channel and leave. And reporting on the closing day means nobody ever learns what the campaign actually produced.
The channel that participants land in matters as much as the campaign itself. Our guide to running a DeFi Telegram group covers the admission control and moderation that decide whether new arrivals stay.
Running a crypto airdrop campaign week by week
The week before a crypto airdrop campaign opens is where most of the outcome is decided. Freeze the task ladder, write down the verification rule for each task in one sentence, and name the person who owns the reward ledger. Take a snapshot of channel size, active posters and wallet activity on the day before opening, because without that baseline every number produced later is a claim rather than a measurement.
During the opening days, read completion by task rather than total signups. If the first task is completing at close to full rate within hours, it is filtering nobody and the ladder needs a costlier rung before the budget drains. If the second task collapses, the cost is priced wrong for the reward on offer. Both readings are available early enough to act on, and almost nobody looks.
The middle of the campaign belongs to the duration tasks and to the channel people landed in. Hold requirements only mean something if someone checks them at intervals rather than once at the end, so schedule the checks and record each result. Meanwhile the channel needs a content rhythm running the whole time, because arrivals who find silence leave before the reward is even distributed.
Closing week is operational. Publish the snapshot rule before the snapshot, so nobody argues about the cutoff afterwards. Decide how rewards are delivered, whether claims expire, and what happens to unclaimed allocations. Then announce the date on which retention will be measured and stick to it, because a number promised in advance is much harder to quietly replace with the closing-day figure.
What to put in the supplier contract
Put the counting moment in the contract in plain words: a participant counts when the action is verified, and verification means this specific evidence. Add what happens when verification fails, whether the supplier replaces those participants or refunds them, and who decides. A supplier who negotiates hard on that clause is telling you where their numbers come from, which is useful before signing rather than after.
Structure payment so that part of it depends on the figure you actually care about. A portion on delivery of verified participants, the remainder released after the retention measurement, keeps both sides pointed at the same outcome. Require a participant-level export as a deliverable, not a summary slide, because a summary cannot be checked and an export can be matched against your own records line by line.
Tracking that survives an audit
Log one row per participant per task: the identifier, the task, the timestamp, the evidence that verification passed, and the source that brought them. That last field is the one projects skip and later wish they had, because a campaign running across several suppliers and channels cannot be attributed at all without it, and every supplier will claim the same good cohort as theirs.
Reconcile the platform count against your own before accepting any report. Deduplicate across wallets and social handles, since the same person entering twice is ordinary and a farm entering dozens of times is not. Keep the pre-campaign baseline beside the final figures so growth is stated as a difference rather than as a total, which is the form in which it can be compared with the next campaign.
Define presence before measuring retention, then never change the definition. Still in the channel, still holding, still posting, or having done anything at all in the window are different questions with different answers, and a project that switches between them across campaigns has no series worth reading. Write the definition into the reporting template and use the same one every time afterward.
When the campaign goes wrong
If the ladder is filtering nobody, raise the cost of the next rung rather than the one people already completed. Announce the change, honor everything earned under the old rules, and explain why in a sentence. Participants accept a tightening that is stated openly and punish one applied silently, and the difference in how the community reads it costs nothing to get right.
When farming is visible mid-campaign, pause distribution rather than the campaign. Hold the suspicious cohort for manual review, keep the rest moving, and publish the rule you applied rather than the list of addresses you rejected. Publishing the rule lets honest participants check themselves against it, while publishing the list starts an argument with people who have nothing better to do than argue.
Flooded channels are a moderation problem before they are a marketing problem. Slow admission when arrivals outpace the people available to answer them, and accept a smaller daily intake in exchange for a channel that reads as alive. The alternative is a room where the only visible messages are reward questions, which tells every genuine arrival what kind of place they walked into.
Plan the edge cases before distribution day. Some participants sit in jurisdictions the distribution cannot serve, some wallets cannot receive the asset, some addresses are contracts that handle it badly, and some people will claim nothing until the deadline has passed. Decide in advance what happens in each case, write it into the campaign terms, and keep the unclaimed pool accounted for rather than quietly absorbed.
If a supplier is behind on delivery near the close, do not extend the campaign to rescue the number. Extending changes the cohort, breaks the comparison with the baseline, and hands the supplier a second attempt at the same tasks with a tired audience. Accept the smaller figure, record what was delivered against what was contracted, and use that record in the next negotiation.
Frequently asked questions
How long should a task campaign run?
Long enough that duration-based tasks are meaningful, which in practice means weeks rather than days. Campaigns compressed into 48 hours can only use tasks that are completable in 48 hours, and those are exactly the tasks that filter nobody.
Do airdrops still work in 2026?
They work as distribution and rarely as retention. A retroactive airdrop to addresses that already used the protocol rewards behavior you wanted; an airdrop designed to attract new addresses mostly attracts addresses created to receive it.
Should rewards be tokens or something else?
Tokens create sell pressure at the moment of distribution, which is worth planning for. Access, allowlist positions, and status inside the community carry no immediate sell pressure and select for participants who value the project rather than the payout.
How do you tell whether a supplier is delivering real participants?
Ask when a participant enters the count, ask for retention measured 30 days after close, and ask what happens to the number if verification fails. Suppliers delivering real participants answer all three without hesitation.
Where to go next
If the campaign is part of a launch rather than a standalone push, DeFi marketing strategy sets out how community work sits alongside PR and paid channels, and KOL-led work covers the alternative of borrowing an existing audience rather than assembling one.
Planning a campaign? Send the brief on Telegram at https://t.me/flexe_io_agency and we will come back with a task ladder and a retention target within 24 hours.
Nothing here is financial advice. Token distribution carries regulatory implications that vary by jurisdiction; take local advice before designing a distribution.