Picture the week after your launch. The campaign worked, wallets connected, the numbers spiked, and the team posted the screenshot. Now try something simple and reach one of those users. Not all of them, just the one person who deposited on Tuesday and hasn't come back. In most protocols we look at, there is no way to do it. You have a wallet address, a transaction record, and silence.

That silence is expensive, because getting the user was expensive. Industry data puts the cost of one activated wallet at around $550, and most apps keep fewer than 10% of their users past 30 days. It gets worse. Research based on Dune's wallet data found that 70% of new wallet users make exactly one transaction and never return, and only 13% of new Binance Web3 Wallet users come back after a week. So this is what actually happens after launch week. You paid for a crowd, most of the crowd left, and you have no way to invite anyone back. So you run another campaign and pay to acquire many of the same people all over again.

Every user you can't reach again is a user you'll pay to acquire a second time.

The second visit is where every business lives

Think about how normal companies work. The moment you sign up anywhere, they get a way to contact you, whether that's an email, a phone number, or a push notification. They do it because the first sale is expensive and every sale after that is nearly free, but only if they can reach you. It's why email still makes $10 to $36 for every dollar spent long after everyone declared it dead. It's the one channel a company actually owns.

Crypto got this backwards. A protocol can see everything its users do, down to the exact deposit, the exact swap, and the exact day someone's balance hit zero. No normal business has ever had that kind of visibility. But it can't say a single word to any of those people. The wallet address is anonymous by design, there's no inbox attached to it, and the only voice most teams have is a brand account shouting into a social feed. You can literally watch a user walk out the door in real time, and you can't even say wait.

The channels that actually reach a wallet

Telegram is the closest thing crypto has to an inbox. It now has over a billion monthly users, 450 million of them daily, and nearly a third have touched crypto. It's also a place to build real products rather than just chats, since around 30,000 mini apps already run inside Telegram, used by 400 million people a month. If you capture a user's Telegram handle at signup, your product, your community, and your "come back" message all live in the same app they open every day anyway.

Messaging a wallet directly is now a real thing. Coinbase Wallet added encrypted messaging back in 2023, built on a network called XMTP, and that network has since grown up. It handled 228 million messages and 15 million registered identities within seven months. This matters because it's the only channel where the thing you already have, the wallet address, is also the contact.

Email still works if you earn it. The catch is that spam filters hate crypto words. Average inbox delivery sits around 83.5% across all industries, and crypto senders land below that. So don't blast people. Ask for the email right after the user gets something good from you, write like a product update instead of a promotion, and keep the list clean.

Your own app is free and nobody uses it. When someone is on your site with their wallet connected, that's the most interested they will ever be in you. Most protocols show that person a generic dashboard. A simple prompt based on what their wallet just did, something like "you deposited, here's the step that makes it earn," is marketing that costs nothing to deliver.

What we'd tell a founder

Ask for a contact at the first win, not the front door. An email box on your landing page collects tourists. The same ask, made right after a successful deposit or first swap, gets you real users. And make the ask specific, because "get notified when your position needs attention" beats "subscribe to updates" every single time.

Let the blockchain tell you when to message. You already know the trigger moments. The balance went to zero, the position sat idle for ten days, or someone bridged in and then did nothing. Connect those events to a message. Web2 teams spend years trying to reconstruct this data from clicks, and you have it for free, in real time. [ADD: one sentence on a real trigger-to-message flow we've run, and what it did to week-four retention.]

Write your first five messages before launch day. That means the welcome, the confirmation, the day-three nudge, the day-fourteen win-back, and the one that goes out when something changes in the product. If these don't exist when you launch, your new users will be gone before you get around to writing them.

Count returns, not sends. One number tells you whether any of this is working, and it's how many wallets came back and did something after you contacted them, compared to wallets you didn't contact. That's the whole scoreboard.

Stop paying for the same user twice

The last cycle trained everyone to treat each launch as a brand-new hunt, with a new campaign, new rewards, a new crowd, and the same collapse at the end. The teams pulling ahead in 2026 figured out something simpler. The most valuable user in crypto is the one who already showed up, and the cheapest growth is bringing that person back. The blockchain tells you exactly who left and when, and the tools to reach them exist today. The only missing piece is deciding that a wallet is the start of a relationship rather than a number on a launch-day chart.