Web3
Web3 product design is mostly an onboarding problem
On-chain products lose 20–40% during onboarding and another 35–50% at the first transaction. What actually removes that friction, and what removing it costs you.
Published benchmarks for on-chain products put drop-off at 20–40% during onboarding and 35–50% at activation, the first real transaction. Compounded, a product can lose more than half the people who had already decided they wanted it.
That isn't a marketing problem. Every one of them arrived, understood the pitch, and clicked. They left during the part we designed.
The first ninety seconds ask more than any other product does
Count what a new user absorbs before receiving any value. Install an extension or an app. Record twelve words they must never lose and never paste anywhere. Fund the account with a currency they may not own yet. Approve a connection request written in vocabulary that exists nowhere outside this industry. Then sign a transaction whose consequences they cannot read and cannot reverse.
A normal SaaS product asks for an email address.
The gap between those two experiences is the entire design brief. Everything downstream; the dashboard, the charts, the position flows; only matters if anyone is still there to see it.
Most of that friction is inherited, not required
This is the part that genuinely changed. EIP-7702 shipped with Ethereum's Pectra upgrade in 2025, letting an ordinary address temporarily delegate execution to a smart contract without migrating to a new one. Sitting on top of ERC-4337's bundler and paymaster infrastructure, it makes four things routine.
A passkey can replace the seed phrase. A paymaster can cover gas, so nobody has to acquire ETH before spending a dollar. Batching collapses approve-then-swap into one signature instead of two. A session key grants a dapp limited authority for a defined window, so the wallet stops interrupting every action.
Each removes a step that was treated for years as the unavoidable cost of being on-chain. None of them ask the user to understand anything new.
The question is which friction to remove, not whether you can
Abstraction has a bill attached, and it arrives late.
Hide custody completely and users won't register that they hold something irreversible until they send it to the wrong address, at which point they open a support ticket you cannot resolve. Sponsor gas indefinitely and you've hidden a running cost from the user and from your own unit economics at the same time. Scope a session key too broadly and you've traded a confirmation prompt for a category of exploit.
The judgment is about where someone genuinely needs to feel the weight of what they're doing, and where the prompt is only ceremony. Most products get this backwards. They interrupt constantly during low-stakes actions, then present the one irreversible transfer in the same modal, in the same tone, as everything else.
Measure the funnel as steps, not as one event
Most teams we meet track "wallet connected" as a single number. It is at least four: opened the connect modal, chose a wallet, approved the connection, signed the first message. Each drops off for a different reason. Until they are instrumented separately, any fix is a guess.
The most expensive part wasn’t adding more features, it was making an already information-heavy trading experience easier to navigate without hiding the data advanced users needed. We simplified the onboarding, order-entry flow, leverage controls, and data hierarchy so users could move from market discovery to execution with fewer steps and much less cognitive load, making the platform feel faster and more approachable without sacrificing depth.
Let people see the thing before they hold the thing
The change that moves activation most is also the least technical: give the product a read-only state that is genuinely useful before any wallet appears. Let someone browse the market, model a position, see real numbers. Ask for the wallet at the moment they want to act, not at the door.
That reorders the funnel so the connection request arrives after the value is obvious instead of before it. People who complete a first transaction within their first day retain at roughly three times the rate of those who don't, and the fastest way to get them there is to stop charging admission at the entrance.
