The Recovery Attempt Log
Write down what you have ruled out, so you never spend a second afternoon testing it.
People who get their wallets back are rarely the ones who tried hardest. They are the ones who kept track. A recovery attempt is a search through a set of hypotheses, and an attempt that finds nothing has still done something valuable: it has removed one. That only holds if you wrote down what it covered.
Without a log, six months of effort collapses into "I tried loads of things", which cannot be reasoned about, cannot be handed to anybody helping you, and quietly guarantees you will run the same six-hour search twice.
Set the target once
Everything below is a search for one specific thing: a phrase that produces this address. Write it once, at the top, and confirm it on a block explorer you navigated to yourself before spending a single hour.
A public address is not a secret, and a phrase is not an address. The address is already on the ledger; writing it down, or giving it to somebody helping you, gives away nothing at all. The words are the entire wallet, and there is no line for them anywhere on this sheet.
The order worth working in
Cheapest and most likely first. The early steps take minutes; the exhaustive ones take hours, and there is no reason to spend those hours before the fast hypotheses are eliminated.
- Restore the phrase exactly as written, in a second standards-compliant wallet, with the derivation path options exposed. A zero balance is often the software looking in the wrong place.
- If there was a passphrase, treat it as the prime suspect before anything else. It is case-sensitive, it has no checksum, and one wrong character silently opens a different, empty wallet.
- Then ordering, using the patterns a phrase actually gets written down in: down the columns of a grid instead of across the rows, snaked along alternate lines, halves swapped, reversed.
- Then one wrong or misread word, across the plausible orderings.
- Then two misread words, which is where handwriting cases usually land.
- Only then the exhaustive searches. They are hours rather than minutes, and they belong last, not first.
Attempt log
One row per attempt. The scope column is the one that pays for the sheet: "tried some orderings" rules nothing out, while "every ordering, with the passphrase exactly as written" eliminates an entire hypothesis permanently.
| Date | What I assumed was wrong | Tool or wallet used | Scope actually covered | Result |
|---|---|---|---|---|
Derivation paths checked
Kept separate from the log above, because it answers a different question: not whether the words are right, but whether the wallet is even looking in the right place. One seed produces completely different addresses on different paths and address types.
| Coin | Derivation path | Accounts and indexes tried | Address type | Result |
|---|---|---|---|---|
What the log is telling you
- An exhaustive re-order that completed with no match proves something exact: with those words and that passphrase, no ordering produces that address. Ordering is now eliminated.
- Which means the fault is elsewhere — a word that is not the word you think it is, or a passphrase that is not character-exact.
- A phrase that restores but shows nothing is a path question, not a key question, until several paths and address types have been checked.
- A word that is not on the standard 2,048-word list is definitely wrong, and the candidates for what it should have been are strongly constrained by both spelling and the checksum.
A completed search that finds nothing feels like a wasted day and is not one. It is a hypothesis eliminated, and it points directly at the next one. That is the entire method, and it is why the log is worth more than the effort.
Educational content, not financial advice. This worksheet structures a decision; it does not make one, and nothing in it is a recommendation to buy anything. We are not a broker, exchange, custodian or adviser, and we never take custody of your assets. We will never ask for a seed phrase or private key. See the full disclaimer.