Free tool

Claims printed on images count too

A badge that says “Certified Sustainable” is a claim, and point 2a of Annex I to Directive 2005/29/EC treats a self-made label as a blacklisted practice. Upload the image and we read what is printed on it.

Choose a label, banner or packshot PNG, JPEG, WebP, GIF, BMP or TIFF · up to 4 MB

We read the text printed on the image. The file is processed in memory and not stored.

Market for the exposure estimate Annual turnover (€) Deeper rewrites (1–2 min)

Free, no account. Rate limited to 5 site scans per hour per address.

These tools need JavaScript to talk to the scanner. Everything else on the site — the 92 restricted terms, the example report and the country guides — works without it.

Why images matter

The badge is usually the worst offender on the page

Text claims tend to be written by someone who has at least thought about them. Badges get designed once and then live on every product page for years.

Self-made labels are blacklisted

Point 2a of Annex I to Directive 2005/29/EC (as amended by Directive 2024/825) bans displaying a sustainability label that is not based on a certification scheme or established by public authorities. There is no misleadingness test to argue about — displaying it is the breach.

Nobody audits the graphics

Copy gets reviewed before launch. A green tick baked into a product template three years ago gets reviewed when someone complains.

Text tools cannot see pixels

Our own website scanner reads HTML text and alt attributes. A claim that exists only as pixels is invisible to it — which is exactly the gap this tool fills.

Honest limits

What OCR cannot give you

Optical character recognition is good at flat, high-contrast type and unreliable at everything else. Script fonts, text over busy photography, tiny legal print and rotated labels all degrade the result.

If the tool returns nothing, that is at least as likely to mean the text was unreadable as that the image was clean. The report shows what it managed to extract so you can judge which happened.

And the visual claim itself — the leaf, the green gradient, the meadow — is outside what any text tool can assess. Assume the imagery is part of the claim, because a regulator will.

Common questions

What does it actually read?

Text printed in the image — badge wording, packaging lines, banner headlines, label copy. It runs optical character recognition and then checks the extracted words with the same lexicon as the copy checker. One difference you should know about: the reader itself is trained on English only, so it will mangle accented characters and non-Latin scripts. German, French and Estonian label text often comes back with the diacritics wrong, and a term that depends on one will be missed. Paste that copy into the copy checker instead.

Can it judge the picture itself?

No, and this matters. A meadow background, a green colour scheme or a leaf motif can carry an implied environmental claim, and regulators do assess implied claims. Reading imagery is not something a lexicon can do — that part stays human.

Which formats work?

PNG, JPEG, WebP, GIF, BMP and TIFF, up to 4 MB. The file type is detected from the bytes, not the name. Sharp, high-contrast text reads best; a photograph of a label at an angle will read worse than a flat export.

Is the file stored?

Partly, and you should know exactly how. The matching happens on our own server, in memory, and nothing is written to disk. But when a match is found, that one sentence — up to 320 characters, or 400 when we draft a rewrite — is sent to a language model behind a private gateway we operate. It keeps request counts, not request bodies, and nothing is used for training. Text with no match is never sent.

Check the copy around the badge too

Most product pages carry the same claim in text and in the image.

Open the copy checker Scan the whole site