Image Metadata: What It Is, How to Check It and When to Remove It

What is image metadata (EXIF, IPTC and XMP)?

Also called embedded metadata or image file metadata, it is data about a picture that travels inside the same file as the pixels. A JPEG keeps its EXIF block in the APP1 segment, PNG and WebP files can carry EXIF too, and IPTC and XMP packets sit in blocks of their own.

A phone photo typically records the camera make and model, shutter speed, aperture, ISO speed, focal length, the date taken, an orientation flag and a geotag when location is switched on. Editors add the software name, an image caption, keywords, a creator field and a copyright notice.

Browsers ignore nearly all of it. They draw the pixels, honor the orientation tag and sometimes apply the color profile, so the rest is dead weight. The IPTC photo metadata overview lists the full field set.

EXIF: what the camera writes

EXIF metadata, short for Exchangeable image file format, was created for digital cameras in the 1990s and is maintained today by the Japanese industry bodies JEITA and CIPA.

A phone held upright often saves the sensor data sideways and records a rotation instruction instead of turning the pixels, so deleting that tag carelessly leaves the portrait photo on its side.

IPTC: the rights and publishing fields

IPTC metadata was designed by the International Press Telecommunications Council for newsrooms and picture agencies. Its descriptive group holds headline, caption and keywords, the rights group holds creator, copyright notice and licensing terms, and the administrative group holds dates and job identifiers.

XMP and ICC: containers and colour

XMP, the Extensible Metadata Platform, is an XML packet from Adobe that can carry IPTC fields and the editing history written by Photoshop or Lightroom. Raw workflows often park the same XMP metadata in a separate XMP sidecar file. An ICC profile describes the color space, such as sRGB, so dropping it from a wide gamut photo can dull the colors.

Does image metadata affect SEO and page speed?

Directly, very little. The Google image metadata documentation names a short list of IPTC fields that Google reads for image license features: creator, Credit Line, copyright notice, digital source type, web statement of rights and licensor URL. Camera EXIF such as ISO, aperture and model is not listed as a ranking signal, so stuffing keywords into them has no documented backing.

The same page says removal can reduce file size and help pages load faster, warns that stripping may be illegal in certain jurisdictions, and recommends keeping the rights and identification fields, such as creator, Credit Line (the attribution line) and copyright notice.

What moves image rankings is visible material: file names, surrounding text, structured data and the attribute covered in our image alt text guide. Metadata is housekeeping, not a ranking lever.

Page weight and load speed

Every embedded block is bytes that each visitor downloads and never uses. The saving per photo is usually small, and we haven't measured a typical share across real sites, so we won't quote one. Compression, resizing and a modern format save far more. See how web page size adds up.

Privacy and safety

This is the stronger reason to care. Phones and GPS cameras can record latitude, longitude and time when location is enabled, and some cameras add a body serial number. A photo of a home office or client site, published as is, can reveal more than the page shows, so remove GPS from photos before upload.

Google Images and licence data

The Licensable badge in Google Images needs a license URL, supplied either as structured data on the page or inside the image through the IPTC Web Statement of Rights field. The Licensor URL field points to a page explaining how to get a license. Keep these fields on pictures you license.

How SEO Site Checker tests image metadata

The wording on this row reads like a measurement, so here is what really happens.

The rule: The Image Metadata Test row is a reminder. It looks at the first 15 images found in your rendered page, never opens their contents, and shows an amber warning on every report.

What the report reads

Our report loads your page in a real browser, collects every image with a non empty src, makes relative addresses absolute and keeps the first 15. Each gets a lightweight headers only request with a short timeout. Those responses feed the image caching and broken image counters, and the modern format check reads the file extension in the URL.

No image body is downloaded. That means no EXIF, IPTC, XMP or ICC data is parsed and no ratio of metadata to file size is calculated. An image that only has a lazy load data attribute and never gets a real src is not seen at all.

What the row shows today

The row has no pass or fail logic tied to your files, so it shows an amber warning for every site, including pages with no images. The red How to Fix button beside the amber icon opens the steps on this page.

It never moves your overall score, and cleaning your images will not turn it green today. The row's own wording talks about images with large metadata, but nothing in the check measures size or compares it to the file, so read that sentence as a prompt, not a finding.

How to use the result

Treat the amber row as a nudge to look inside a few of your own files, because the report is not a metadata checker. Start with hero images, team photos, phone shot product photos and anything from a client location.

Check my website

How to check the metadata of an image

To check image metadata, start with the viewer your computer already has, then use ExifTool for the rest, and keep the original untouched while you look.

1. Get the exact file. Save the image your page really serves, because a CDN or plugin may have changed it from your original.
2. Use the built in viewer. On Windows, open the file's Properties and read the Details tab. On a Mac, open the file in Preview and choose Tools, then Show Inspector. Neither lists everything, but both reveal GPS and camera fields.
3. Run ExifTool for the complete list. It can read EXIF data, IPTC and XMP from nearly any image, and printing the group name beside each tag shows which block holds it.
4. Write down what you don't want public. Typical candidates are GPS coordinates, serial numbers, the owner name, software and thumbnails.

exiftool -a -G1 -s photo.jpg
exiftool -a -G1 -s '-GPS*' photo.jpg

The first command lists every tag with its family name in front. The second prints only tags starting with GPS, whichever family stores them, because location can sit in EXIF or in XMP. If it prints nothing, no such tags were found. In the Windows Command Prompt, use double quotes around "-GPS*" instead of single quotes.

Get ExifTool from exiftool.org, with brew install exiftool on a Mac using Homebrew or the libimage-exiftool-perl package on Debian and Ubuntu.

Browser developer tools show file size and headers but not EXIF. An online EXIF viewer may upload your file to a server or read it inside your browser, so check its privacy statement and keep private photos out.

How to remove image metadata from website images

The order matters more than the tool. Follow it to strip EXIF data without the two classic accidents, sideways photos and lost copyright.

1. Rotate first. If a photo depends on the EXIF orientation tag, bake the rotation into the pixels, or use the rotate and strip command below. Otherwise portraits turn sideways.
2. Work on copies. Several commands overwrite files, and deleted metadata can't be recovered. jpegoptim also overwrites its input by default.
3. Choose the level. Strip everything from decorative images. For pictures you own, license or deliver to clients, keep the copyright and creator fields.
4. Re-upload under a new name. Images are cached for a long time, so a new filename or CDN purge delivers the clean file, as the image caching guide explains.

Strip with ExifTool or jpegoptim

ExifTool is the reference tool, and the ExifTool documentation covers every flag below. Use the first command for decorative images and the second for photos you own or license, which wipes the file and writes the rights fields back from the original. The third repeats the wipe across a folder, which is how you batch remove metadata, so run it on a copy. ExifTool edits the metadata blocks without recompressing pixels, so JPEG quality is untouched.

# 1. Wipe one JPEG in place
exiftool -all= -overwrite_original photo.jpg

# 2. Wipe, then copy the rights fields back
exiftool -all= -tagsFromFile @ -Copyright -Artist -IPTC:CopyrightNotice -IPTC:By-line -IPTC:Credit -XMP-dc:Creator -XMP-dc:Rights -XMP-photoshop:Credit -XMP-xmpRights:WebStatement -XMP-plus:LicensorURL -overwrite_original photo.jpg

# 3. A whole folder (point at the folder, not a wildcard)
exiftool -r -ext jpg -ext jpeg -ext png -ext webp -all= -overwrite_original photos/

# 4. jpegoptim strips and optimizes the JPEG losslessly, in place
jpegoptim --strip-all photo.jpg

# 5. Apply the orientation tag, then strip (ImageMagick 7; version 6 calls it convert)
magick photo.jpg -auto-orient -strip clean.jpg

In command 2 the IPTC:Credit and XMP-photoshop:Credit tags hold the Credit Line field Google reads, and adding -Orientation -icc_profile to the copy list keeps those too. The wipe also removes the digital source type field, and C2PA content credentials will be removed or invalidated, because their signature no longer matches an edited file. For AI generated images, add the XMP-iptcExt:DigitalSourceType tag to the copy list and test it on a copy first.

A full strip also removes the ICC profile, so ExifTool warns that colors may change. Command 4 is handy because it drops comments as well and usually shrinks the file, but jpegoptim only touches JPEG, so use ExifTool (command 3) for PNG and WebP. Command 5 fixes sideways portraits in one step, which saves a separate rotate pass, although ImageMagick re-encodes the JPEG and quality drops slightly, so prefer ExifTool or jpegoptim when the pixels should stay as they are.

Strip at export in your editor or optimizer

Lightroom Classic's Export dialog has a Metadata panel that runs from all metadata down to copyright only, plus a Remove Location Info checkbox. Photoshop's Export As dialog offers a shorter Metadata choice, typically none or copyright and contact info. Labels change between versions, and the copyright and contact option publishes your details, so check before exporting.

Our free image compressor usually drops EXIF and color profile data when it shrinks a file. Rotate portrait photos upright first, because the tool removes the orientation tag without applying it. Then confirm with ExifTool, because when it can't make a file smaller it returns the original with its metadata untouched.

Without the command line

On Windows, open the file's Properties, go to the Details tab and click Remove Properties and Personal Information. On a Mac, the Preview inspector's GPS tab has a Remove Location Info button. Both clear only part of the data, so check the result.

WordPress, Shopify and Wix

Don't assume your platform strips metadata. WordPress keeps your original upload in the media library, so a photo with GPS stays downloadable at its original address, while the smaller sizes it generates may or may not keep metadata. Strip before upload, then run ExifTool on a served URL. For Shopify and Wix, test one served file.

Confirm the result

Run the same check on the cleaned file and confirm the GPS, camera and software lines are gone and portraits stand upright.

What image metadata to keep and what to remove

Decide once for the whole site and write the rule into your upload workflow.

• Keep rights fields. On images you own or license, keep creator, copyright notice, Credit Line, web statement of rights and licensor URL, the IPTC fields Google reads.
• Keep or bake in orientation and color. Either keep the orientation tag or rotate the pixels first. Convert wide gamut photos to sRGB before dropping their color profile.
• Remove private data. GPS position, device serial number, owner name, software and editing history, comments and the embedded thumbnail do nothing for visitors.
• Think about provenance. For AI generated or edited images, keep digital source type and C2PA content credentials if you want provenance to travel with the file. A blanket strip removes or breaks both.

Common image metadata mistakes

Lost ownership: Deleting copyright and creator fields on licensed or client work removes evidence you may need later, and Google reads the same fields.

• Sideways portraits. Stripping the orientation tag leaves a phone photo lying on its side, so rotate the pixels first.
• The old GPS pin is still downloadable. A re-upload under the same name leaves the CDN serving its cached copy, so use a new filename.
• GPS survived on a platform you trusted. Some networks and CMS uploads remove location and others pass it through, so test one real file.
• Keyword stuffed EXIF fields. Writing search terms into camera tags changes nothing, because Google doesn't document camera EXIF as a ranking signal.
• Private photos in random viewers. Some online tools upload the file to a server, so keep client pictures out.