Here is the document. A phone photo of a printed trade price list - tilted, unevenly lit, shot the way anyone actually photographs a piece of paper that's lying on a desk:

Look at what a machine has to survive to turn this into a spreadsheet. The page is rotated a few degrees. The lighting changes across the sheet. The table has category bands ("COMPOSTS & MULCH", "TOOLS & SUNDRIES") interleaved with product rows. Units are inconsistent - "50 L bag", "each", "3-pack", "pair". Two price columns sit side by side, and above the table there's a header block with an address, phone extensions, and delivery terms that belong in a summary, not in the rows.

This is Image to Excel, a free tool: paste or drop that photo, and this comes back:

Two sheets, 24 rows. The category bands became a column instead of garbage rows. The prices are numbers, not text - you can sum them the moment the file opens. The header block became a main sheet of field and value pairs: company name, effective date, minimum order, delivery radius. Nothing was retyped.

The tool is built paste-first. Take a screenshot, press Ctrl+V (Cmd+V on a Mac) anywhere on the page, and extraction starts on its own - no file dialog, no convert button. That covers the most common way tabular data gets trapped in a picture in the first place: a dashboard that won't export, a table inside a PDF viewer, a webinar slide, a price list someone sent over WhatsApp.

Dropping or choosing a file works too - PNG, JPG, WEBP, TIFF, and HEIC photos straight from an iPhone (HEIC converts to JPG in your browser before anything uploads).

Free OCR sites read the characters and hand you a wall of text. On the photo above, that means the row structure, the two price columns, and the category bands are all your problem again - you've traded retyping for reformatting. The extraction here reads the layout itself: which value belongs to which row and column, which lines are headers, which text is a field rather than a cell. Every table in the image becomes its own sheet, and anything outside the tables lands on a main sheet.

Rotation and skew are corrected before extraction, so the tilt in the photo above doesn't tilt the rows.

The page ships with three fictional samples - a crisp SaaS billing screenshot, the tilted price list photo above, and a mobile banking app screenshot - so you can run the whole flow without touching your own data. Each one renders its result instantly, and real uploads run through the same live pipeline.

Free, no signup. Images up to 14MB, with a daily number of free runs per network. Files travel encrypted, are processed on SOC 2 and ISO 27001 certified, HIPAA compliant infrastructure, are encrypted at rest, and are never used to train models. The samples are fictional.

If you're photographing or screenshotting the same report on a schedule, the retyping you're avoiding is a workflow you could delete entirely. DocuPipe - the platform this tool runs on - ingests recurring documents automatically, any format, and returns clean data under a schema you control. That story is in why we're doubling down on free tools.