Common data-entry errors in bankruptcy petitions — and how to catch them

Most defective consumer bankruptcy petitions are not the product of bad lawyering. They are the product of transcription. A number was written by hand, read under time pressure, and typed into a field — and somewhere in that chain a 4 became a 9, a creditor name became an abbreviation that cannot support notice, or a monthly figure landed in the wrong column. The legal analysis was fine. The data was wrong before the analysis touched it.

The cost of a data error is not proportional to its size. A one-digit error in a Social Security number and one in a furniture valuation are the same keystroke, but one is harmless and the other forces an amendment. This guide sorts the errors by type, traces what each costs, and sets out a review structure that catches them at the cheapest point.

A taxonomy of petition data-entry errors

Transposed and misread digits

The most common error in the workflow is a digit that changed value between the page and the screen. Transposition happens at the keyboard — the eye reads 4831, the fingers type 4813. Misreading happens at the page: a handwritten 1 with a serif reads as 7, an open 4 as 9, a closed 0 with a stray mark as 6, a hurried 3 as 8.

These land in Social Security numbers on Form 121, account numbers throughout Schedules D, E, and F, and every dollar figure on the schedules. They are dangerous because the output looks perfectly normal — nothing about “4813” signals that it should have been “4831.”

Names that break creditor noticing

Creditor names fail in two directions. Clients write shorthand — “Cap One,” “the hospital,” “Dr. Patel” — and a staffer enters it as written, producing a matrix entry that does not identify the noticing entity. Or a name is misspelled, which is the same failure: the notice goes out and does not arrive. Servicers compound this, because the entity a client pays is often not the one entitled to notice. Building a defensible list is covered in our guide to the creditor matrix.

Amounts in the wrong place

Column and unit errors are quieter than digit errors but often larger in magnitude: gross pay entered where net pay belongs on Schedule I; a biweekly figure entered as monthly; a balance entered as a monthly payment; secured, priority, and unsecured amounts on the wrong schedule.

Missing creditors and stale addresses

Omission has the longest tail. Clients forget medical accounts already in collections, co-signed obligations, family loans, and old accounts that have changed hands twice. Addresses go stale between the last statement a client kept and the filing date, especially for collection agencies. An old address and a wrong one produce the same result.

Cross-form inconsistency

Some errors are visible only when two forms are read together. Income on Schedule I that does not reconcile with Form 122A-1 is the classic case: the two use different definitions and lookback periods, so they legitimately differ, but the difference has to be explainable. Similar mismatches appear between Schedule A/B and Schedule D collateral, and between Form 107 and the schedules.

What each error type costs

The pattern is consistent: each is far cheaper to fix before filing. Pre-filing, a correction is a keystroke. Post-filing, it is an amended schedule, a supplemental notice, and — if the trustee has engaged — an explanation.

Why single-person re-keying has an irreducible error rate

It is tempting to treat these as a training problem. They are not, or not only. Manual transcription has a floor that attention cannot push to zero, for structural reasons.

• Reading handwriting is inference, not perception. An ambiguous glyph is resolved by expectation, and expectation is often wrong. The reader feels certain either way.
• The error is silent. A transposed digit produces a well-formed number, so there is no feedback signal at the moment the mistake is made.
• Self-checking is weak. Re-reading your own entry re-activates what you intended to type, which is why it catches far less than a second person’s read.
• Fatigue is cumulative. Accuracy on field 400 of the day is not accuracy on field 40, and intake is a high-field-count task.
• Interruption resets context. A call mid-schedule reliably produces a skipped or duplicated line.

The conclusion is not that people are unreliable. It is that one unverified read of any field is not enough, and no amount of care converts one read into two. Accuracy has to come from the structure of the process, not the effort inside it.

Layered prevention that actually works

1. Verify identity fields against a source document. Social Security number from the card or a tax return; legal name exactly as it appears on government ID.
2. Reconcile creditors against the credit report line by line. It is the closest thing to a complete list, and the pass surfaces omissions and name mismatches at once.
3. Enter by source document, not by screen. Finish the credit report, then the questionnaire, then the collection letters.
4. Run a fixed cross-form consistency pass. Schedule I against Form 122A-1 and the pay stubs, Schedule A/B against Schedule D, Form 107 against the schedules. A petition review checklist makes it repeatable.
5. Give high-consequence fields a second read by a second person — identity fields, creditor names and addresses, and the largest amounts, not the whole petition.
6. Keep the source document beside the entry. Digit errors are caught instantly when the handwriting sits next to the typed value, and almost never when verification means walking to a folder.

That last point is the one most systems get wrong. Verification that requires retrieving paper does not happen at volume, so the check that matters most is skipped most often. Software can close the gap by displaying a cropped image of the handwriting beside each captured value. Casewell is built around that pattern — several independent reads per field, disagreement surfaced for staff confirmation, then a verified export into Best Case. The principle holds regardless of tooling: a field two independent reads agree on is stronger than one a single person typed.

None of this eliminates review — it concentrates it. Time saved re-reading correctly captured fields is time available for the judgment calls that require it, an argument developed in the bankruptcy paralegal workflow. Moving data into petition software once and cleanly is the point of importing client data into Best Case. The forms these entries feed are published at uscourts.gov.

This guide is general information for law-firm staff, not legal advice for any particular case.