Three hidden addresses from a florist's contact page
This is the kind of snippet you find in the source of a small shop's site: one protected link, one protected span and a footer link that also carries a subject line. On screen all three read as [email protected] or Email us.
The decoder finds all three, lists each address once, and keeps the full decoded text where it differs from the address, so you can see the footer link was set up for stockist enquiries.
<p>Orders: <a href="/cdn-cgi/l/email-protection" class="__cf_email__" data-cfemail="3c534e58594e4f7c545d4e5e53494e5e505353514f1259445d514c5059">[email protected]</a></p> <p>Weddings: <span class="__cf_email__" data-cfemail="7e091b1a1a1710190d3e161f0c1c110b0c1c121111130d501b061f130e121b">[email protected]</span></p> <footer><a href="/cdn-cgi/l/email-protection#91f9f4fdfdfed1f9f0e3f3fee4e3f3fdfefefce2bff4e9f0fce1fdf4aee2e4f3fbf4f2e5acc2e5fef2faf8e2e5b1f4ffe0e4f8e3e8">Email us</a></footer>
[email protected] data-cfemail exact match [email protected] data-cfemail exact match [email protected] protection link [email protected]?subject=Stockist enquiry
Decoding a page
Open the page source
On the business's contact page, right-click and choose View page source (or press Ctrl+U in most desktop browsers). The protected addresses only exist in the source, not in text you copy from the screen.
Paste it in
Copy the whole source, or just the lines around the placeholder, into the box. A single link or a column of hex strings works too.
Read the results table
Each row shows the address, where it was found (attribute, protection link or bare hex) and either exact match or the full decoded text. Rows that could not be decoded say why, for example a string that was cut short when you copied it.
Copy, download or rebuild the source
Copy addresses gives one per line. Download CSV saves address, decoded text, source type, hex and any note. Show decoded source returns your paste with each protected element replaced by the plain address, handy if you are feeding HTML into another tool.
Where the encoded string hides
Cloudflare uses a few patterns, and the decoder looks for all of them in the same paste.
| What you see in the source | Where it comes from | What the decoder does |
|---|---|---|
data-cfemail="3c53..." on a span or link with class __cf_email__ | An address that was written as visible text on the page | Decodes the attribute value |
href="/cdn-cgi/l/email-protection#91f9..." | A mailto link; the part after # is the encoded target, which can include ?subject= | Decodes the part after #, keeps the subject in the decoded text column |
href="/cdn-cgi/l/email-protection" with no # | A link whose address sits in a data-cfemail attribute on the same element | Reads the attribute instead |
| A line of hex on its own | You copied only the attribute value | Treats any run of six or more hex digits as a candidate when no attribute or link is present |
How the scrambling works, worked by hand
The scheme is a single-byte XOR, which is why it can be undone without any key from Cloudflare. Read the hex two characters at a time. The first pair is the key; every pair after it is one byte of the address XORed with that key.
Take the first example above. The string starts 3c 53 4e 58. The key is 3c. Then 53 XOR 3c is 6f, the letter o; 4e XOR 3c is 72, r; 58 XOR 3c is 64, d. Carry on to the end and you get [email protected]. A different page, or the same page tomorrow, may use a different key, so the same address can appear as two unrelated-looking strings. That is why Copy addresses and the counts work on the decoded result rather than the hex: two strings that decode to the same address count once.
The bytes are read as UTF-8, so accented characters survive. If they do not form valid text, the row says the string was damaged rather than guessing.
When decoding is not the whole story
Decoding tells you what the site published, not whether the mailbox still works or who reads it. Small business sites often keep an old footer address long after the mailbox closed.
Some things this tool cannot do: it does not visit the site for you, so you need the source yourself; it cannot read addresses that a site only shows as an image; and it does not help with pages that hide contact details behind a form, a login or a CAPTCHA. Those are access controls, and they should be respected.
Before you add decoded addresses to a list
- Keep the page URL next to each address so you can show where it came from.
- Drop privacy, abuse, unsubscribe and no-reply inboxes; they are not outreach contacts.
- Check the domain still receives mail (an MX lookup) and verify the mailbox before a first send.
- Decide whether a named address belongs to a sole trader or a partnership, because UK electronic marketing rules treat them like individuals.
- Respect the reason the owner switched protection on: collect the business's published contact details only, at a gentle pace.
Questions people ask
How do I decode a Cloudflare protected email address?
Copy the page source, find the data-cfemail value or the text after email-protection# and paste it here. The first two hex digits are a key, and each following pair is XORed with that key to give one character. The decoder does that for every string it finds and lists the resulting addresses.
Why does a website say [email protected] instead of an email?
The site is served through Cloudflare with Email Address Obfuscation enabled. Cloudflare replaces each address in the HTML with an encoded string and a placeholder, then a script restores it in visitors' browsers. Tools that read raw HTML, and copies of the source, only see the placeholder until the string is decoded.
Is decoding Cloudflare email protection legal?
The decoder only reverses an encoding of an address the site already shows to every visitor, and it runs on text you supply. What matters legally is what you do next: UK GDPR and PECR still govern how you store and use the address. This page is not legal advice; the ICO's guidance on electronic marketing is the place to check.
Can I decode many addresses at once?
Yes. Paste a whole page, several pages of source, or a list of hex strings one per line. Every attribute and protection link in the paste is decoded in one pass, duplicates are removed, and the CSV holds every row. The tool stops at 5,000 encoded strings per paste to keep the tab responsive.
Why does a row say the string looks cut short?
Each character takes two hex digits, so a valid string always has an even number of them. An odd count usually means the copy stopped one character early or picked up a stray letter. Copy the attribute again from the page source, making sure you take everything between the quote marks.
Will Map Extract's own crawler see these addresses?
Map Extract's email extractor is built to handle common address obfuscation on the pages it crawls and records the page each address came from. Coverage depends on how a site hides its addresses, so check the source column if an address you expected is missing.
Where your data goes
The decoding is a few lines of JavaScript that run in this tab. The page source you paste and the addresses it contains are not uploaded or logged anywhere, and they disappear when you reload or close the page. The CSV file is built on your own device.
Checked against the tool on