How to convert image to Excel comes down to four real methods: Excel's built-in Data from Picture, the Microsoft 365 mobile app, an online OCR converter, or asking ChatGPT to read the table. Each turns a photo of a table into editable cells; which you pick depends on your Excel version, device, and how sensitive the data is.
None of these is our own tool's job, and I will say so plainly below: a compressor moves bytes around, it does not read a table out of pixels. What this site does do is get a messy source photo into shape before you hand it to one of these readers. For the other conversions people run here, the image conversion hub maps them all.
How to Convert Image to Excel: The Four Methods at a Glance
There is no single button that does this everywhere, so the honest answer is a short menu. One reads the table straight into cells but needs a current Microsoft 365 license; two more upload your picture to somebody else's server; the last one is a chatbot that guesses well but has to be checked. Here is the whole decision in one table.
| Method | Reads table structure | What you need | Best when |
|---|---|---|---|
| Data from Picture | Yes, straight into cells | Microsoft 365, build 15723 or later | A clean, head-on shot and a modern Excel |
| Microsoft 365 mobile app | Yes, straight into cells | The Microsoft 365 phone app | The table is on paper in front of you |
| Online OCR converter | Sometimes, layout can slip | A browser and a file to upload | No Microsoft 365, and data you are fine uploading |
| ChatGPT | Sometimes, always review | A ChatGPT account | Odd layouts a strict reader chokes on |
Read the table top to bottom: the first two methods are Microsoft's own and keep your data on your account, the bottom two hand your image to a third party. That privacy line matters more than accuracy for anything with names, card numbers, or salaries on it, and I come back to it near the end. If you have a modern Excel, start at Method 1 and only drop down the list when it fails you.
Does our tool convert an image to Excel?
No. This tool compresses and converts between JPG, PNG, WebP, and AVIF. Reading a table's structure out of a photograph and rebuilding it as spreadsheet cells is optical character recognition, the same job Adobe Acrobat and Google's OCR engines do, and a compressor is not that kind of software. Anyone who tells you a plain image compressor exports XLSX is selling you something.
Where this site actually helps is the step before conversion. When the data lives on paper, you photograph it, and phone photos are heavy. A single page shot straight off the camera runs about 4.5 MB, and a three-page report is roughly 13.5 MB before you have imported a single row. Run each page through the compressor here first and that same set drops to about 2.7 MB, near 900 KB a page, an 80 percent cut, with the row and column edges Data from Picture depends on still sharp because quality 80 stays visually lossless on photographs.
| Capture | Size before | Size after |
|---|---|---|
| One page photo | 4.5 MB | 0.9 MB |
| Three-page report | 13.5 MB | 2.7 MB |
If your source is already a screenshot it is probably a PNG, and compression here is lossless on PNG, so every pixel stays exactly as captured. Cropping out anything that is not the table is still your job, and it is the single biggest thing you can do for a clean read.
Method 1: Excel's built-in Data from Picture
Data from Picture is the most direct route when you have it: it reads table structure, not just characters, and drops the result straight into cells instead of a wall of text. Microsoft describes it as a way to "turn images that have table data into data that you can edit in Excel" (Microsoft Support). It is the Excel Data from Picture feature everyone means when they ask how to do this at all.
Picture From File vs Picture From Clipboard
There are two ways to get the image into the dialog. For a file you have already saved, click Data, then From Picture, then Picture From File, and select it. For a table you have not saved anywhere, screenshot it first with Windows logo key plus Shift plus S, then choose Data, From Picture, Picture From Clipboard instead (Microsoft Support). Either way the dialog shows Excel analyzing the image, then hands you a review screen. Check the recognized values against the source, fix anything wrong by hand, and click Insert Data. Do not skip that review. Excel expects corrections before you commit, and a misread 8 as a 3 in a column of totals will not announce itself later.
Requirements and where to find the button
This is where most people get stuck. Data from Picture needs Excel for Microsoft 365, version 2210, build 15723, or later (Microsoft Tech Community); it does not exist in Excel 2016, 2019, or 2021, no matter how current your license feels (Microsoft Learn). It runs on Windows, the web, macOS, and iOS or Android, but the Windows desktop app has its own floor: Windows 11, or Windows 10 version 1903 or later, with the Microsoft Edge WebView2 Runtime installed. If the From Picture button is missing from your Data tab, you can surface it through File, Options, Customize Ribbon, though that only shows the button and will not add the feature to a version that lacks it. Recognition also covers a fixed list of about twenty mostly European character sets, so a table in Japanese, Arabic, or Russian will not read correctly regardless of photo quality.
Method 2: Scan paper with the Microsoft 365 mobile app
When the data is printed on a physical page, your phone is the fastest scanner you own. Open the Microsoft 365 or standalone Excel app on iOS or Android, tap Actions in the bottom right, and choose Image to Table. Take a picture of the table, using the cropping handles to isolate just the data, then tap Confirm to process it. You get the same review step as the desktop, so edit any questionable values before you save. Send the file to your email or drop it straight into OneDrive, and it opens on your computer as a normal workbook.
This method shares the desktop feature's engine and its limits, so the same language list and the same head-on, well-lit framing rules apply. The advantage is purely physical: a phone held over a page beats trying to photograph paper with one device and import it on another.
One page at a time is the rule that trips people up. Image to Table reads a single table per shot, so a multi-page report becomes a series of scans you stack in the same workbook, not one giant capture. Keep each shot to one page and one table, confirm each read as you go, and paste them into adjacent sheets or below one another rather than trying to make the phone swallow the whole document in a single frame.
Method 3: Online OCR converters and the free manual route
No Microsoft 365? A web-based image to excel converter will do the job in a browser: you visit the site, upload your JPG or PNG, choose Excel as the output, and download the resulting spreadsheet. These tools are genuinely convenient for a one-off, and the good ones handle common table layouts well. The catch is that you are uploading your image to a company's server to be processed, which is fine for a public price list and a poor idea for a payroll sheet.
Online image to excel converter tools versus plain OCR
There is also a free route you already have. Upload the image to Google Drive, right click it, choose Open with, then Google Docs, and Drive runs OCR automatically into a new document. Copy that text into Excel and use Data, Text to Columns to split it by a delimiter. Budget real time, though: Google's own documentation warns that "tables ... are not likely to be detected," so the grid usually does not survive and you rebuild rows and columns yourself (Google Drive Help). That is the difference between a table-aware converter and plain text OCR, and it is why a dedicated tool is worth it when the layout matters. If your source happens to be a JPG, any of these will convert JPG to Excel or to editable text; the format of the picture is not the constraint, the tool's table logic is.
Method 4: Can ChatGPT convert an image to Excel?
Yes, with review. Upload the image to ChatGPT and ask it, in plain words, to read the table and return it as rows and columns you can paste into Excel, or as a downloadable file. It is the most forgiving method for awkward layouts, merged headers, or a receipt a strict reader would refuse, because a language model infers structure instead of matching a rigid grid. That same flexibility is the risk: it can quietly invent a plausible value where the image is unclear, so you check every number against the picture, not just the ones it flags.
Three habits make the read far better, and they are the same habits Data from Picture rewards. Give it a high-quality, in-focus image; crop tight to just the table so there is nothing else to interpret; and prefer standard typed fonts over handwriting or decorative type. Then read the output against the source before you trust a single total.
How accurate is image to excel conversion?
Accuracy is good, not perfect, and the errors cluster in predictable places, which means you can review for them on purpose instead of hoping. Every method above, Microsoft's included, expects you to correct the read before you commit it. The failures are almost never random; they are the handful of confusions any character reader makes.
| Error pattern | What you see | Where it bites |
|---|---|---|
| 0 read as O, 1 as l | a number turns into a letter | ID codes, SKUs, part numbers |
| Decimal drift | 1,234.5 splits or merges wrong | currency and totals columns |
| Merged cells | two columns collapse into one | spanned headers, grouped labels |
| Dropped rows | a faint or shadowed row is skipped | long tables, photocopied pages |
Three things drive those errors more than anything: resolution, angle, and lighting. Microsoft's own guidance says to keep the shot head-on and focused and to crop out everything that is not the table, because perspective distortion and clutter are top causes of a bad read (Microsoft Support). Even lighting matters more than raw megapixels past a point, since a shadow across half the table can read as a false column border. Shoot straight, light it flat, crop tight, and review the numbers.
Is it safe to upload financial data to these tools?
It depends entirely on which method you choose, and this is the decision most guides skip. Excel's Data from Picture and the Microsoft 365 mobile app process the image inside your own Microsoft account. An online OCR converter or a public chatbot, by contrast, means uploading a picture of your data to a third party's servers to be processed, and you are trusting their retention and privacy terms with whatever is in that table.
For a screenshot of a public spreadsheet, nobody cares. For a bank statement, a payroll table, or anything with names attached to numbers, keep it on Microsoft's native features and off the web uploaders. If you do compress the source here first, note that we strip EXIF, the hidden metadata your camera writes into a file including GPS coordinates, by default, and browser uploads are deleted within one hour; but the safest data is the data you never handed to a stranger's server in the first place.
When should you not convert an image to Excel?
When the data is not actually trapped in a picture. If the table already lives in a real spreadsheet, a CSV, or a PDF with a working text layer, skip every method above and export or copy it directly; you will get a perfect result with zero misreads. Image conversion exists for when a photograph or screenshot is genuinely the only copy you have.
Skip it too when the language is outside the supported list, where a manual transcription is the honest fix rather than a cleaner photo, and when the data is sensitive enough that no third-party upload is acceptable and you do not have a native Microsoft option. In those cases, typing it in is slower but it is also correct and private, and sometimes that is the trade you want.