Someone sends you a price list as a screenshot. Or a bank statement as a phone photo. The numbers are right there, but you can't sort them, sum them, or paste them into anything useful. So you retype them, which is slow and exactly how 4,500 turns into 5,400.
I built Toolghar's Image to Excel tool for this situation. This post covers how it works, how to get a clean result on the first try, and what to fix afterwards, because a bit of cleanup is normal.
#What the converter actually does
Two things happen. First, OCR reads the text in the image. Second, the tool works out the table structure: where each row starts, which column each value belongs to, and which row is the header.
The second part is the one that matters. Plain OCR gives you a block of text with numbers in roughly the right order. A table-aware converter puts values in the right cells, so column C is actually prices and not a mix of prices and product names.
With the Image to Excel converter you get an .xlsx file with one sheet. The first row holds the column headers it detected, and each row below is one row from your table. Cell colors, bold text, and borders aren't carried over. You get the data, not the styling.
#A worked example
Say a supplier sends you this as a screenshot:
| Item | Unit | Price (₹) |
|---|---|---|
| A4 paper, 500 sheets | ream | 320 |
| Blue ballpoint pen | box of 10 | 95 |
| Stapler, medium | piece | 180 |
After conversion you get a sheet with Item, Unit, and Price (₹) in row 1 and the three products underneath. Add a Qty column in D, put =C2*D2 in E2, fill down, and you have an order total in a minute.
Three rows aren't hard to type. Sixty rows are, and that's where this pays off.
#Step by step
#1. Start with the best source you have
If the table exists digitally, take a screenshot instead of photographing your monitor. Zoom in first so the text is large. For paper:
- Lay the page flat and shoot straight down, not at an angle.
- Use even light. Glare on glossy paper can wipe out whole cells.
- Tap to focus on the table so small print stays sharp.
#2. Crop to the table
Cut out logos, addresses, footnotes, and anything else that isn't the table. Text around the grid can turn into stray rows. Any photo editor works, or use the Image Cropper.
If a page has two separate tables, crop them into two images. One table per image gives much cleaner output.
#3. Upload
The tool accepts JPG, PNG, and PDF, up to 12 files at once and 5 MB per file. If only a few pages of a long PDF have the tables you need, pull them out first with Split PDF, or turn them into images with PDF to Image. A focused input converts more reliably than a 40-page report.
#4. Convert and download
Each file is processed separately and gets its own .xlsx download. It opens in Excel, Google Sheets, LibreOffice, or Numbers.
#5. Check it against the source
Put the original next to the spreadsheet. If the table has a total row, the fastest check is a =SUM() on the column compared with the printed total. If they match, you're almost certainly fine. If they don't, the difference usually points straight at the misread cell.
WARNING
[!WARNING] For financial, tax, or medical data, treat the output as a first draft and reconcile it against the source before relying on it.
#Common problems and how to fix them
Most issues fall into a handful of buckets, and each fix takes a minute or less.
#Numbers are stored as text
They sit left-aligned, sometimes with a small green triangle, and SUM returns 0. Select the column, go to Data > Text to Columns, and click Finish. That makes Excel re-read the values as numbers. =VALUE(A2) in a helper column also works.
#Decimal and thousands separators are swapped
A European statement writes 1.250,50 where an Indian or US one writes 1,250.50. If the source uses a different convention from your Excel settings, fix it with Find and Replace, or change the separators under File > Options > Advanced.
#Negative numbers in brackets
Accounting statements often show (1,200) instead of -1200, and that can come through as text. On that column, replace "(" with "-" and ")" with nothing, then convert to numbers.
#Merged cells and multi-level headers
If a header like "Q1" spans three columns with Jan, Feb, and Mar underneath, the output only has one header row. Rename the columns to something like "Q1 Jan", "Q1 Feb", "Q1 Mar". On complex tables this is the cleanup you'll do most often.
#Two values in one cell
Sometimes a date and a description land in the same cell. Use Text to Columns with a delimiter, or type the split you want in the next column and press Ctrl+E (Flash Fill) so Excel copies the pattern down.
#Look-alike characters
0 and O, 1 and l, 5 and S, 8 and 3 on a smudged print. Rare on clean screenshots, more common on photos of faded thermal receipts. The SUM check catches most of them.
#Leading zeros disappear
Product codes like 00457 can lose their zeros if Excel treats them as numbers. Format the column as Text and fix those cells, or rebuild them with =TEXT(A2,"00000").
#Rows missing at the edges
If the first or last row is gone, the crop was probably too tight or the edge was blurry. Retake it with a small margin around the table.
TIP
[!TIP] If a photo keeps fighting you, retake it. Thirty seconds with better light beats fixing twenty cells by hand.
#Where it works well and where it struggles
| Source | What to expect |
|---|---|
| Screenshot of a web or app table | Very clean, little or no cleanup |
| Scanned PDF of a printed report | Clean if the scan is sharp and straight |
| Phone photo of a printed page | Good with flat paper and even light |
| Bank or card statement | Good, but check signs, separators, and dates |
| Faded thermal receipt | Hit or miss, mostly because of the print quality |
| Handwritten table | Works on neat writing in a clear grid, needs proofreading |
Tables without visible grid lines often work too, as long as the columns are clearly aligned. Accuracy drops on very dense tables with tiny text. For those, zoom in and take two screenshots (top half, bottom half) instead of one.
#After the data is in a spreadsheet
From here it's normal Excel work: sort, filter, pivot tables, charts. If the data needs to go somewhere else:
- Excel to JSON if you're loading it into an app or an API.
- JSON to Excel for the reverse, like turning an API response into a sheet.
- CSV to JSON if you exported the sheet as CSV along the way.
#What it costs
Toolghar gives you free daily credits, which cover the occasional screenshot or statement. For a big batch there are optional one-time credit packs. No subscription required. Details are on the pricing page.
#Quick recap
- Screenshots beat photos, and flat, sharp, well-lit photos beat angled ones.
- Crop to one table per image before uploading.
- Expect a flat sheet with one header row. Merged headers need renaming.
- Compare the column total to the source, especially when it's money.
- Text-as-numbers, swapped separators, and bracketed negatives are the usual fixes.
Next time a table arrives as a picture, run it through the Image to Excel converter before you start typing. For other format jobs, the full list is on the tools page.