# Low Quality Scan to Excel: When to Rescan Instead

- Author: FastBankConvert (https://fastbankconvert.com)
- Published: 2026-10-17
- Category: Scanned and image statements
- Canonical: https://fastbankconvert.com/blog/low-quality-scan-to-excel

> Can a low quality scan go to Excel? The signs a scan is too poor for OCR, what each flaw costs in errors, and when a fresh scan is the faster fix.
A low quality scan can go to Excel, but every flaw in the image turns into wrong or missing cells you then have to find. Blur and low resolution swap digits, skew scrambles rows, and faint print drops characters. With the paper at hand, a fresh 300 DPI scan is usually quicker than repairing a poor one.

## Can a low quality scan go to Excel at all?

A low quality scan usually still produces something in Excel, because OCR returns its best guess for every character it can find. The question is not whether you get cells, but how many of them are right. A poor scan gives a sheet that looks complete and hides its errors among the correct values.

That is the real cost of a weak scan. An empty cell is obvious. A cell holding 1,770.00 instead of 1,170.00 looks like any other number, and you find it only by checking. The worse the scan, the more checking each page needs, until checking takes longer than the conversion saved.

So judge the scan before converting, not after. A minute spent looking at the image at full size tells you whether the output will need a light spot-check or a line-by-line review. The six signs in the next section are what to look for.

Not every poor image is a scan problem. A picture of a chart or graph will convert poorly however sharp it is, because the values are shapes rather than text, as [rebuilding the numbers behind a chart image](/blog/chart-image-to-excel) explains.

## Six signs a scan is too poor for OCR

Six visible signs predict poor OCR: blurred characters, a low resolution, a tilted or curved page, faint or uneven print, blocky compression around letters, and text cut off at the page edge. Check for them by zooming to 100 percent over the numbers that matter, because the whole-page view hides most of them.

1. **Blur.** The edges of characters are soft and grey instead of sharp. Common in phone photos taken in poor light or with a moving hand.
2. **Low resolution.** Characters look jagged or made of visible squares, and small digits lose their gaps. Typical of scans at 150 DPI or lower, or of an image shrunk to send by email.
3. **Skew or curve.** Rows of text slope or bend, often near a fold or a book spine.
4. **Faint or uneven print.** Characters are pale, broken or vanish into a grey background. Old thermal paper and worn ribbons both cause this.
5. **Heavy compression.** Small blocks or ripples appear around each letter, a sign the image was saved as a low-quality JPEG.
6. **Cut-off edges.** The scan clips part of the amount or balance column at one side of the page.

One sign in mild form is usually fine. Two or more, or one in severe form over the amounts, is a reason to stop and rescan. The guide to [scanning paper so it ends up in Excel](/blog/scan-to-excel) covers the settings that avoid each of these at the start.

## What each low quality scan flaw costs in spreadsheet errors

Each scan flaw produces its own kind of spreadsheet error, which helps you predict what to check. Blur and low resolution swap similar digits, skew moves values into the wrong row or column, faint print drops characters and signs, compression adds stray marks, and clipping removes whole values.

The table summarizes each flaw, the error it tends to cause in the converted sheet, how you notice it, and the fix.

| Scan flaw | Typical spreadsheet error | How you notice it | Fix |
| --- | --- | --- | --- |
| Blur | Look-alike digits swapped, such as 3 and 8 | Running balance stops adding up | Rescan with the page flat and still |
| Low resolution | Digits merged or misread, decimal points lost | Amounts a hundred times too large | Rescan at 300 DPI, or higher for small print |
| Skew or curve | Values shifted into the row or column next door | Rows with an empty amount and a doubled one below | Straighten before OCR, or rescan flat |
| Faint print | Missing minus signs, broken digits, blank cells | Debits shown as credits; gaps in columns | Rescan in greyscale with a little more contrast |
| Heavy compression | Stray marks read as commas, points or dashes | Amounts that will not convert to numbers | Rescan and save as PDF or high-quality image |
| Cut-off edges | Whole values or the balance column missing | Fewer columns than the page shows | Rescan with the full page inside the frame |

Errors in the amount column are the expensive ones, because they change totals. Errors in descriptions are cheaper, because a person still recognizes "GROCRY MART" as a grocery store. When you decide whether a scan is good enough, look hardest at the numbers.

Credit card statements deserve special care here. They often lack a running balance on each line, so a misread amount cannot be caught row by row, and [converting a scanned credit card statement to Excel](/blog/scanned-credit-card-statement-to-excel) explains how to check them against the printed totals instead.

## Repairing a weak scan image versus scanning again

Repairing a weak scan means editing the image before OCR: straightening, cropping, adjusting contrast or removing noise. It helps with skew, dark borders and mild shadows. It cannot restore detail the scan never captured, so blur, low resolution and heavy compression stay broken however much you edit.

The dividing line is simple. If the information is in the image but obscured, repair can help. If the information is missing from the image, only a new scan brings it back.

Repairs that work:

- **Rotate and straighten** a tilted page so rows run level.
- **Crop** dark borders and anything outside the page.
- **Raise contrast moderately** on a slightly faded but sharp scan.

Repairs that do not work, or make things worse:

- **Sharpening a blurred image.** It adds halos that OCR can read as extra strokes.
- **Enlarging a low-resolution image.** Upscaling makes the squares bigger, not the digits clearer.
- **Heavy contrast on faint print.** Pale strokes either vanish or thicken until neighbouring digits touch.

Time is the other factor. A few repairs on a handful of pages are quick. Repairing forty pages one by one, and then still checking each one closely, is usually slower than feeding them through the scanner again.

## When is a fresh scan the faster fix?

A fresh scan is the faster fix whenever you still have the paper and the scan shows two or more of the six signs, or any sign over the amount column. Rescanning a page takes a minute or two. Finding and correcting misread amounts on that page afterwards often takes much longer, and you may miss some.

A quick decision checklist, to run on one sample page:

1. Zoom to 100 percent over the amounts. Can you read every digit without effort?
2. Are the rows level, with no curve near the edges?
3. Is the whole page, including the right-hand balance column, inside the image?
4. Are minus signs, brackets and decimal points all clearly visible?

Four yeses: convert. Any no on questions 1 or 4: rescan. A no on 2 or 3 alone: try a repair, then convert.

Once you do convert, plan the checking. A long statement with a balance check on every row means many formulas, and on a big workbook Excel recalculates all of them after each edit. If correcting cells starts to lag, [setting Excel calculation to manual for big sheets](/blog/set-excel-calculation-to-manual) keeps the sheet responsive while you work.

FastBankConvert is being built to read scanned statements as well as digital ones, with scans taking longer, and to tell you before download if the balances do not add up, which is exactly where a weak scan shows itself. It is getting ready to open; you can join through [Get started](/get-started).

## Poor scans with no original left to rescan

When the paper is gone and the scan is all you have, the goal shifts from accuracy to known accuracy. Convert the scan, then check it as fully as the document allows, and mark anything you could not confirm. A sheet with clearly flagged doubtful cells is more useful than one that looks clean and is not.

Steps that help:

- **Ask for another copy.** Banks can usually provide past statements, often as digital PDFs that need no OCR. For other documents, the sender may still have the file.
- **Use every built-in check.** Running balances, page totals and summary figures all let you test the amounts by arithmetic.
- **Read the worst areas by eye.** Where the scan is too poor for OCR, a person can often still read the figure with some effort, especially with the neighbouring values for context.
- **Type rather than repair.** For a few badly damaged lines, typing the values from the image is faster than cleaning up OCR output.
- **Record what you could not verify.** A note column saying "illegible on scan" protects anyone who relies on the sheet later.

Keep the poor scan alongside the converted file. If a question comes up about a figure, the image is the only evidence of what the page said. Store it under a name that matches the spreadsheet, so the two stay together when files are moved or shared.