Alt text embedded inside image files via IPTC metadata
Bake descriptive text into the image binary so it survives CDNs, re-hosting, and copying.
What this signal tests
We fetch a sample of your content images and parse their XMP metadata packet (the embedded structured-data block inside JPEG, PNG, and TIFF files). We look for the IPTC fields Iptc4xmpCore:AltTextAccessibility (a short description) and ExtDescrAccessibility (a longer one). At least 20 percent of content images should carry a non-empty AltTextAccessibility field.
Why it matters for your visibility in AI
Alt text in HTML lives only on your page. The moment someone embeds your image elsewhere, screenshots it, or downloads and re-uploads it, the alt text is gone. IPTC AltTextAccessibility is embedded inside the image file itself, so it travels with the bytes. An AI system fetching your photo from anywhere on the web gets authoritative descriptive text directly from the file metadata. This is increasingly important as AI systems ingest images from a wider variety of sources: social media shares, news syndication, academic mirrors, RAG indexes of mixed provenance. The on-page alt text might be wrong, missing, or AI-generated; the embedded metadata reflects what the original publisher actually said. For news photography, stock images, and any content licensed for reuse, IPTC metadata is the only way to ensure attribution and description survive.
Pass criteria at a glance
| Criterion | Passes when |
|---|---|
| >=20% of content images carry non-empty AltTextAccessibility XMP field. |
How we test it
Our scanner downloads the binary of each sampled image and parses its XMP packet using a library like exifr or exiftool. It extracts the Iptc4xmpCore:AltTextAccessibility and ExtDescrAccessibility properties and counts how many images carry non-empty values for the former. We report the percentage and pass when it reaches 20 percent.
Show technical detection method
Fetch image binaries; parse XMP packet for AltTextAccessibility/ExtDescrAccessibility fields.
If your site fails: how to fix it
- In Adobe Photoshop or Lightroom, the field is labelled "Alt Text (Accessibility)" under File Info > IPTC Extension. Photoshop 2021+ added explicit UI for this; older versions need the XMP panel.
- For command-line workflows, use exiftool: `exiftool -XMP-iptcExt:AltTextAccessibility="Red leather armchair photographed from the side" image.jpg`. Script this into your image processing pipeline.
- If you use a digital asset management (DAM) system like Bynder, Brandfolder, or Cloudinary, check whether it exposes AltTextAccessibility in its metadata schema. Most modern DAMs do; older ones need a custom field.
- Update your image upload workflow so the alt text typed into the CMS also writes to the IPTC field on the file. WordPress plugins like Smush and ShortPixel can do this; for custom builds, hook into the upload pipeline.
- Verify CDNs and image processors preserve XMP metadata. Cloudflare Images, ImageKit, and Cloudinary preserve it by default; some optimisation tools strip it to save bytes. If yours does, configure it to preserve XMP.
Quick facts
| Maturity | EMERGING |
|---|---|
| Weight | low |
| Category | Multimodal |
Primary sources
Related signals
Frequently asked questions
How is this different from the HTML alt attribute?
HTML alt is page-bound: it works only when someone is looking at your page. IPTC AltTextAccessibility lives inside the image file, so anyone who downloads, embeds, or syndicates the image still has it. Both should match; both serve different consumers.
Do major social platforms preserve IPTC metadata when I upload?
Sadly, no. Twitter/X, Facebook, and Instagram strip most metadata on upload. LinkedIn preserves more. Adobe Lightroom's publish modules preserve metadata where possible. The metadata still matters for crawler-to-crawler propagation, just less for social shares.
What about ExtDescrAccessibility? Is that important too?
It is the long-form description (think paragraph rather than sentence). Useful for complex images like charts, infographics, and detailed photographs. Where AltTextAccessibility is the elevator pitch, ExtDescrAccessibility is the full description. Both improve AI ingestion when present.
Does adding metadata make my files bigger?
Marginally. A typical XMP packet with alt text adds 200 to 1000 bytes per image. Across a catalogue this is rounding error compared to the image bytes themselves. The benefit clearly outweighs the cost.
Run your own scan
Run a free scan and see how your site grades across all 155 AI-readiness signals.