Crypto marketing is not a “small test”. It is one long funnel experiment.
Rules to avoid false signals:
- do not judge performance in the first 24-48h
- measure 2 horizons: D1 activation and D7 retention (D30 if you can)
- separate market beta with a baseline or holdout
- optimize for cost per retained wallet, not cost per click
- filter sybil/bot traffic before calling it a win or a fail
On-chain actions to track: wallet connect, swap, stake, bridge, net deposits.
What is the most misleading “quick win” you have seen in Web3?
