# OCR Adjuster Notes: Turning a Claim Log Into a Table

- Author: Schema Law (https://www.schemalaw.com/)
- Published: 2026-04-20T16:49:00Z
- Updated: 2026-10-11T19:22:20Z
- Category: Practice areas
- Canonical: https://www.schemalaw.com/blog/ocr-adjuster-notes

> What OCR does with printed adjuster notes: how a claim notes log is laid out, how to turn its entries into a sortable table, and which details to proofread.

Yes. OCR reads printed adjuster notes well, because a claim notes log is typed text from a computer system. Turning those notes into something you can sort takes one more step: each entry has to become a row with its date, author and text in separate columns. Handwritten field notes are a separate and harder problem.

## What do adjuster notes look like on the page?

Printed adjuster notes look like a ledger of diary entries. Each entry has a header line with a date, a time, the author's name or user ID and often a category, followed by a block of free text. A page header repeats the claim number and insured, and entries run on for dozens of pages.

Claim systems differ, but most printouts take one of two shapes.

- **Stacked entries.** The header sits on its own line, sometimes shaded or in bold, and the note text runs the full width of the page beneath it. A rule or a blank line separates one entry from the next.
- **A grid.** Date, author, category and note text sit side by side in columns with ruled lines, and the note column wraps over many lines within one tall row.

Either way the same parts are present, and it helps to name them before you convert.

| Part | Example | Why you want it separate |
| --- | --- | --- |
| Entry date and time | 03/14/2026 09:42 | Sorting and finding gaps |
| Author | TWHITFIELD | Filtering by adjuster or supervisor |
| Category | Coverage, Contact, Reserve, Diary | Filtering by subject |
| Note text | "TC to insd re cause of loss..." | Searching and quoting |
| Page of the file | LC-0027 | Citing the entry |

Two features cause trouble later. Long entries break across pages, with the page header and footer landing in the middle of a sentence. And a printout that runs newest entry first makes the story read backwards. The log is one section of a larger file, which the post on [how to OCR a claim file](https://www.schemalaw.com/blog/ocr-claim-files) covers as a whole.

Category labels help you filter, and a plain search helps you find a bad-faith claim note, a claims-made policy date, or a line about the duty to indemnify. Pull those entries when you review the log for a coverage dispute. Match the date, the author, and the sentence you plan to quote to the printout before the row leaves your sheet.

## Can OCR read a printed claim notes log?

OCR can read a printed claim notes log with few character errors when the print is clean, since the log is uniform system type with no handwriting. What it cannot do reliably without your help is keep the structure: which words are a header, where one entry ends and which column a value belongs to.

Expect these results on a typical page.

- **The words come through.** Sentences in the note text read correctly apart from the usual trouble spots: small type, faint toner and anything that was faxed before it was scanned.
- **A ruled grid is usually recognized as a table.** Printouts with visible column lines tend to export with their columns intact. The post on scanned tables explains [how OCR finds a table on a scanned page](https://www.schemalaw.com/blog/scanned-pdf-to-excel#how-does-ocr-find-a-table-on-a-scanned-page) and why ruled lines help.
- **Stacked entries come out as running text.** With no column lines, the header and the note arrive as consecutive paragraphs. The content is all there, but nothing marks a header as a header.
- **Stamps intrude.** Bates numbers and confidentiality legends at the foot of each page appear in the text wherever the page ended.

That is enough for searching on day one. A searchable PDF of the log lets you find every mention of a name or a policy term at once. Sorting and filtering need the table described below. Test a few pages of your own log before you plan around any particular result, since system printouts vary.

## Turning a claim notes log into a sortable table

Turn a claim notes log into a table by exporting it to a spreadsheet, giving each entry one row, and splitting the header into date, time, author and category columns. Add a column for the page each entry came from. Then sort by date and read the claim in the order it happened.

The route depends on which shape your printout has.

1. **Convert the log on its own.** Cut the notes section out of a copy of the claim file and upload it as a separate document, with its page range in the file name.
2. **Export to Excel or CSV.** In Schema Law both formats are included on every plan. The [export formats listed on the homepage](https://www.schemalaw.com/#formats) show the choices.
3. **For a grid printout, tidy the rows.** Delete the repeated column headings from the top of each page. Where an entry broke across a page, join the two halves into one cell.
4. **For stacked entries, mark the headers.** Add a column that tests whether a line begins with a date, then copy each header's date, time and author down to the lines of text beneath it. Merge those lines into one cell per entry.
5. **Split the header.** Separate date, time, author and category into their own columns, and set the date column to a true date format so it sorts by calendar, not by first digit.
6. **Add the source page.** Record the Bates number or file page for each entry. An entry you cannot cite is of little use.
7. **Number the rows in their original order** before you sort, so you can always restore the carrier's sequence.

A finished row, from an invented water damage claim, looks like this.

| Seq | Date | Time | Author | Category | Note | Page |
| --- | --- | --- | --- | --- | --- | --- |
| 212 | 14 Mar 2026 | 09:42 | TWHITFIELD | Contact | TC to insd re cause of loss. Insd states leak first noticed 2/8. Advised will review plumber rpt. | LC-0027 |

With the table built, filtering does the reading for you. Filter the author column to see only supervisor entries. Filter the category to "Reserve" to list every reserve change by date. Sort by date and scan for a three-week stretch with no entries at all.

## Abbreviations, claim numbers and dollar amounts to proofread in adjuster notes

Proofread three things in converted adjuster notes before you rely on them: the shorthand, the identifying numbers and the money. Each is a place where a recognition error produces something that still looks plausible, so no spell checker or quick skim will catch it.

**Abbreviations.** Adjusters write in clipped shorthand: "insd", "clmt", "TC", "LM", "ROR", "RS", "IA", "DOL". Usage differs by carrier and by person, so build a glossary from the file itself and confirm the meanings with someone who knows that carrier's practice. Recognition has [no dictionary](https://www.schemalaw.com/blog/does-ocr-use-a-dictionary) to lean on here. "clmt" can come out as "cimt", and a lowercase L, a capital I and the digit 1 are easily confused. Search for the misread forms as well as the right ones.

**Claim and policy numbers.** A claim number mixes letters and digits with no pattern a reader can sense-check. Compare it character by character with the first notice of loss or the declarations page. One wrong character in a page header matters little. One in an entry that refers to a companion claim sends you looking for the wrong file.

**Dollar amounts.** Reserve changes, payment amounts, estimate totals and settlement authority all appear in the notes as bare figures. Check every amount you intend to cite against the image, looking at the decimal point, the thousands comma and any minus sign or parentheses. Where the notes record a payment, find the matching check or payment screen elsewhere in the file and see that the two agree.

**Dates inside the text.** The entry date in the header says when the note was written. A date inside the note says when something happened. "Insd states leak first noticed 2/8" was typed on 14 March. Keep the two apart in your table, and add an "event date" column if the difference matters to your timeline.

## Handwritten adjuster notes and field worksheets

Handwritten adjuster notes and field worksheets are the pages software struggles with. Results on handwriting vary from usable to useless depending on the hand, the pen and the scan. Plan to read these pages yourself and to type out the parts you need, treating any recognized text as a rough guide.

They turn up in the file as scanned attachments more often than in the log itself.

- **Scope sheets.** Room dimensions, material notes and quantities jotted during an inspection, often on a clipboard form.
- **Roof and floor sketches.** Diagrams with measurements written along the edges. The drawing is not text, and the measurements sit at angles.
- **Notes from a recorded statement.** Jottings made while a claimant or witness was talking.
- **Annotated estimates.** A printed estimate with handwritten changes to quantities or prices.

The post on [what OCR can do with handwritten notes](https://www.schemalaw.com/blog/ocr-handwritten-notes) explains why handwriting behaves so differently from print and how to give a page its best chance.

For the claim file, two habits are enough. List every handwritten page in your index with a note of who appears to have written it and when. Then look in the typed log for an entry that summarizes the same inspection. Where the handwritten scope sheet records a measurement or an observation that the typed entry leaves out, that difference is worth a line in your notes and a question for the adjuster.

## Try it on a real document

Upload a scanned PDF and export it as Word, a searchable PDF or plain text on the free plan. Your documents are never used to train an AI model. [Start free](https://www.schemalaw.com/api/auth/google)