How to convert a screenshot to Excel

Sometimes the data you need only exists as an image — a table in a PDF, a screenshot someone sent you, a photo of a printed report — and retyping it isn't realistic if it's more than a few rows. This is the reverse of screenshotting a spreadsheet: you're starting with a picture and want editable cells out of it. It only works because of OCR (optical character recognition), and it isn't perfect, so knowing where it breaks matters as much as knowing the steps.

Excel's "Insert Data from Picture"

On Windows, Excel has a built-in importer for this: Insert → Data → Data from Picture (also on the Data tab, or from your phone's Excel app camera). Point it at an image file or a live camera shot of a table, and Excel runs OCR, shows you a preview of what it detected, and lets you fix cells before inserting the result onto the sheet.

It's genuinely useful for clean, well-lit, high-contrast tables — printed reports, simple grids. It's currently a Windows and mobile feature; it isn't available in Excel for Mac's desktop app as of this writing.

Where OCR gets it wrong

The failure modes are predictable once you've hit them a few times:

• Merged or nested headers. A two-row header (a category spanning several sub-columns) usually flattens incorrectly, since OCR reads left-to-right without understanding the visual grouping.
• Similar characters. 0/O, 1/l/I, and commas vs. periods in numbers are the most common misreads, especially at lower image resolution.
• Low contrast or small text. A screenshot taken at low resolution, or a photo with glare, drops accuracy fast — this is the single biggest factor, more than table complexity.
• Currency and percentage symbols. These sometimes get dropped or misplaced relative to the number, so totals can look right but be off by a factor of 100.

The practical takeaway: always treat the import as a draft, and spot-check the numeric columns against the source image before trusting the sheet. For anything with more than a handful of ambiguous cells, it's often faster to retype the table by hand than to fix OCR errors one at a time.

Third-party OCR tools

If the image isn't a clean table (a photo taken at an angle, a scanned page with a table embedded in other text), dedicated OCR services generally handle more variation than Excel's built-in tool, at the cost of an extra export/import step to get the result into a workbook. They're worth reaching for when Excel's importer can't find the table boundaries at all.

The other direction

If you're doing the reverse — you already have the data in Excel and need it as a clean image instead — that doesn't need OCR at all. Celtrim lets you upload a .xlsx, .csv, or .tsv file, select the range you want, and export it directly as a PNG, so the round trip (image → Excel → image) never has to touch a lossy OCR step in the second half.