Nobody Has Ever Wanted A Smart Account.

Somebody is going to tell you this quarter that your users should be on smart accounts.

They will have good reasons and you should take them seriously. A person signing up to your product should not have to write twelve words on a piece of paper and keep them safe for the rest of their life. They should not have to go and buy a currency they have never heard of, on an exchange they have never used, before they can press one button in your app. Both of those lose you real customers, and anybody who has watched their own signup funnel fall apart already knows it.

Then you will ask what it costs. The answer comes back in engineering quarters, and you will probably approve it, because the only other thing on the table is doing nothing. What nobody puts in the proposal is a list of what the account actually does once your users are on it.

That list can be written from the public record. Every action taken through a smart account is announced by the account itself the moment it happens, where anybody can read it. We read three months of Ethereum and a fortnight of Base, 6,507,481 actions, and every one of them falls into one of two things.

Nothing falls outside those two. About half of everything done through one of these accounts is several steps landing together instead of one at a time, and more than half of it was paid for by a company rather than by the person acting.

The comparison you have been handed is a fair one, and nobody in it is selling snake oil.

Coinbase Smart Wallet really does remove the piece of paper, which is the single biggest reason ordinary people bounce off this whole category. Privy and Dynamic really do get somebody who has never heard of a wallet all the way through a signup with an email address. ZeroDev, Biconomy and Alchemy really do let you pay your users' fees for them. Safe does the same job for the company's own money rather than its customers'.

And it works at scale. On Base, in one fortnight, 905,300 accounts took 4,519,908 actions, which is 322,851 a day, and 1.44% of them failed. Every name on the shortlist clears the bar, so picking between them is not where your money goes.

On Ethereum, across three months, 65.5% of every smart account action was paid for by somebody other than the person taking it. On Base it is 52.3%. Put both chains together and it is 3,666,953 actions out of 6,507,481.

Most of that is a handful of consumer apps that decided their users should never see a fee, and are paying to make it so. On Base there are 82 companies doing it. One of them covers 41.6% of it by itself.

Five companies account for 86.8% of every action anybody paid for on behalf of somebody else. The other 77 share what is left. When somebody tells you the industry is moving to sponsored transactions, these five are most of what they mean.

The fee itself is tiny. On Base, all 4,519,908 actions cost 40.54 ether in total, about $100,000 at $2,478 an ether, which works out at 2.2 cents an action. On Ethereum it is 21 cents.

The largest sponsor on Base covered 983,303 actions in the fortnight. At the chain's average of 2.2 cents, that is roughly $22,000 a fortnight, or about $570,000 a year. That figure is modelled rather than measured: it applies the whole chain's average fee to that one company's action count, because the record shows who paid and how often, but not what each company was billed.

So a company is spending something like half a million dollars a year to spare its users a charge of about two cents. Nobody does that to save their users two cents. They do it because, without somebody else paying, a new user cannot take their first action until they have left, opened an account somewhere else, bought a currency and come back, and most of them never come back. The money buys the removal of that errand.

There have been three generations of this machinery, one in 2023, one in 2024 and one in 2025. Companies migrating today are being pointed at the newest.

On two chains, over two separately measured windows, the 2025 generation fails four to five times more often than the one it replaces. It is being adopted anyway, and it already carries 27.8% of Ethereum's actions.

None of the vendors on your shortlist did anything wrong here. The machinery is still changing underneath them, and whichever version your engineers pick this quarter is the version somebody on your team has to keep up with next quarter.

Write the two behaviours down and the proposal reads differently.

You want several steps to land as one, so a person is not clicking through four screens and abandoning at the third. You want your company to pay the fee, so nobody has to buy a currency to use your product. Those are the two things 6,507,481 actions were spent on, and they are what you are approving a migration for.

A new account for every user you have is one route to the first of them. It costs engineering quarters, a second class of user you support for as long as you exist, and an awkward year in which half your customers are on one kind of account and half are on the other. Some of you will pick it anyway and be right to. You should pick it having seen what else gets you there.

The habit of building something new on the chain to get a behaviour has a track record we can read, too. In two separate samples of Ethereum, 934 contracts were deployed directly by a person or a script, and when we read each one's whole life back, 512 of them had never left a public record of doing anything. Plenty of those were tests, scripts and bots rather than anybody's product, so it is not a verdict on any one of them. It is what new things put on the chain mostly turn into.

Plug gets you several steps landing as one without moving a single user and without deploying anything: programmable Ethereum transactions, without smart contracts.

The biggest sponsor on Base is modelled at about $570,000 a year to cover a fee of 2.2 cents. That is what one company thought it was worth to have its users never leave the app. Before you spend a quarter of engineering on the account, price the behaviour you actually wanted.