When a transaction is not worth the fee
Three dollars is a rounding error on a three thousand dollar move. On a twenty dollar move it is fifteen percent, and you pay it twice, because eventually you have to get back out.
That is the whole subject. Everything below is that one sentence applied to the places where it actually bites.
I run mobile proxy lines and a rack of Android phones for a living, so a lot of my week goes on costing things that look free. A SIM is about ten dollars a month. A modem depreciates at roughly a dollar fifty a month across its useful life. A powered hub carrying thirty ports is a hundred and fifty dollars, and the machine everything hangs off was two hundred. I know those to the dollar, because a line I cannot cost is a line I might be selling at a loss for a year without noticing.
Onchain fees never get that treatment. They are small one at a time and no invoice turns up at the end of the month, so they never pile up anywhere you can look at them.
Fees are floors
The price of an action is set by the network, and it is roughly flat. A swap costs about the same on a twenty dollar trade as on a twenty thousand dollar one, because the work the chain does is identical either way.
Flat costs behave in a way people are bad at intuiting. As a proportion of the amount they get worse the smaller you go, and they get worse quickly.
Below some size a position is trapped. Nothing is locking it. It is worth less than the transaction required to move it, so it stays where it is. The wallet still displays the number, and the number is real, it just cannot be reached unless the chain gets cheaper than it was on the day you funded it.
Nothing warns you when you cross that line, because you cross it at funding time, months before you find out.
Count the round trip
Almost everybody budgets one transaction for what turns out to be six.
Getting into a position on a chain you are not already on: a bridge, then an approval, then the deposit or the swap. Three fees before the thing you meant to do has happened.
Then the activity itself, which is rarely a single action.
Then the exit. A withdrawal, a swap back, a bridge home.
The exit half is the one worth thinking about, because you pay it later at a price nobody can read today. If the chain is quiet when you leave, fine. If it is busy you pay busy prices, and you pay them at the exact moment you least want to, since the usual reason for leaving in a hurry is that something has changed.
So the honest cost of a position is the entry, plus the activity, plus an exit fee you can only estimate. Estimate it high.
Waiting costs nothing
Fees on a busy chain swing by a factor of five or ten inside a single day. Same contract, same call, same outcome, different price depending on what else is competing for blockspace that hour.
Very little of what people do onchain has a genuine deadline. A deposit you meant to make this week does not care whether it lands on Tuesday night or Sunday morning.
I keep a list of transactions waiting for a quiet hour. Some have been sitting on it for weeks and none of them deteriorated.
It is the only lever in this activity that is free to pull, and it is the one people skip, because clicking now feels like progress and clicking later feels like nothing.
Be suspicious of deadlines generally. Manufactured urgency is the standard method for getting somebody to pay peak fees on a decision they have not finished making.
Bridging, and arriving with nothing to pay with
The quoted bridge fee is one line item out of three or four.
There is the send on the origin chain. There is whatever the bridge itself takes. There is a fee on the destination side before you can do anything at all. And if that value ever comes home, the entire sequence runs again backwards.
Then the trap that catches nearly everyone once. You bridge a stablecoin to a chain you have not used before. It arrives, the balance is correct, and you cannot touch it, because fees on that chain are paid in that chain’s own native token and you have none of it. You own the balance. You cannot move the balance.
The fix is a second transfer of a small amount of native token, out of a wallet that already has some, at its own fee. Five minutes if you saw it coming. A stuck afternoon if you did not.
The reverse version is quieter and worse. You spend the native balance down to almost nothing, leave an amount too small to cover one more transaction, and now everything else on that wallet sits behind a fee you cannot pay while the wallet shows a perfectly healthy total.
The rule I use: never let the gas balance on any wallet fall below about three times what a bad hour costs on that chain. It is dead capital and it is the cheapest insurance available anywhere in this hobby.
Cleaning up costs money too
Approvals are transactions. Revoking them is a transaction as well, one per token, per spender, per chain.
A wallet carrying twenty stale approvals is twenty transactions to clear. On a cheap chain that is small change and you should go and do it. On an expensive one during a busy hour, the cleanup can cost more than the tokens those approvals cover.
That is the real reason stale approvals sit on wallets for a year. Somebody ran the numbers and disliked the answer.
Two things follow. Timing, first: the revoke sweep belongs on the waiting list with everything else, run on a quiet weekend instead of on the afternoon it finally occurs to you. Placement, second: I will connect an experimental wallet to something on a chain where cleanup is nearly free, and I will not on a chain where every mistake carries a removal fee.
That second one is a preference plenty of people will argue with. It means I do less on expensive chains than the opportunity there might justify. I would rather keep the ability to clean up than the chance to be early, and I am aware that costs me things.
Consolidate on a quiet week
Anybody running several wallets eventually wants the value sitting in fewer places.
The right week to do that is one where nothing is happening. The week most people actually do it is the one where something has happened and the funds are needed immediately, which is when the chain is expensive and waiting is off the table.
The second reason to do it early is dust. Every wallet you fund becomes a future leftover balance, individually too small to bother with, collectively worth more than you would enjoy donating. Sweep them while sweeping is cheap and most of them survive. Leave it, and the chain keeps them.
What I got wrong
I funded eight wallets on a busy chain at roughly forty dollars each, because the activity I wanted was there and I was not thinking past the deposit.
I had never costed the round trip. Forty dollars into a wallet that needs an approval, a deposit, a withdrawal and a bridge home, on a chain where each of those can run several dollars on an ordinary day, leaves very little on the other side even when nothing goes wrong.
Three of the eight are below the line now. The money in them is real, it is not coming back, and I can point at the afternoon I made it unrecoverable. The run cost me somewhere around two hundred dollars in total, and I am fairly confident the five that did work returned less than their own fees.
There is a smaller one I like better as a lesson. Seventeen dollars of a token sitting on a wallet I still open occasionally, worth less than the transaction that would move it, created in four seconds by splitting an amount into pieces that were too small for where I was putting them. It is the cheapest thing I own and it is the reason I started counting.
What changed afterwards is one number. Before funding anything now, I work out a full round trip at a bad hour on that chain and multiply by five. If I am not prepared to commit at least that, the wallet does not get created.
That rule has stopped me doing more things than any other rule I keep, which I have decided is the point of it.
What the arithmetic actually says
A meaningful share of small scale activity in this space does not clear its own transaction costs.
Read that as arithmetic. Subtraction holds no opinion about anybody. The costs are certain and payable now, there are more of them than people count, and whatever comes back is uncertain, arrives later if it arrives, and is not something anyone is in a position to promise.
The sums improve at larger sizes, which is uncomfortable, since the people most drawn to this are usually the ones working with the least.
None of this is financial advice, and I am not going to tell you what a token will be worth or which projects deserve your money. What I can tell you is what the transactions cost, because that part is arithmetic, it is knowable in advance, and it is the only half of the equation you actually control.
More cost sheets and tested tooling are on the site.
Get new guides and videos first — join the Telegram channel.