People read web addresses the way they read sentences: left to right, looking for familiar words. The domain name system works the other way. The part that decides who owns a site sits towards the right, just before the top-level domain. Almost every link-based deception exploits that mismatch, or exploits the fact that our eyes forgive small differences that computers treat as completely different names.

Find the registrable domain first

Every hostname has a registrable domain: the name someone actually bought from a registrar, such as example.com or example.co.in. Everything to its left is a subdomain, and the owner of the registrable domain can create any subdomain they like.

So a host that reads yourbank.com.account-verify.example belongs to whoever owns account-verify.example. The words “yourbank.com” at the start are simply a subdomain label chosen to catch the eye. The skill is to find the top-level domain at the right (taking care with two-part endings like .co.in or .gov.in), step one label left, and that is the owner. Only then look at anything else.

A related trick uses the part of a URL before an at-sign. URL syntax allows a username section before the host, and browsers ignore what comes before the at-sign when deciding where to connect. A link that appears to begin with a trusted brand followed by an at-sign and another domain actually goes to that other domain. Modern browsers warn about or strip this, but email clients and chat apps may still display it in full.

The families of lookalike domains

Typosquatting registers common misspellings: a dropped letter, a doubled letter, adjacent keys swapped. These catch people who type addresses by hand as well as people skimming a link.

Character substitution relies on shapes that look alike in many fonts: the digit 1 for a lowercase l, a zero for the letter O, or the pair “rn” for “m”. At normal reading speed, a domain using “rn” in place of “m” is almost invisible.

Homoglyph or IDN attacks go further. Internationalised domain names allow characters from scripts beyond basic Latin, which is important for Hindi, Tamil and many other languages. But some characters in other scripts, such as certain Cyrillic letters, are visually identical to Latin ones. A domain made from those characters can look pixel-perfect identical to a real brand. Underneath, such domains are stored in an ASCII form called Punycode, which begins with the prefix xn--. Major browsers apply rules to display that Punycode form instead of the lookalike characters when scripts are mixed suspiciously, but the protection depends on the browser and the specific characters.

Combosquatting adds a plausible word to a real brand: brand-support, brand-refund, brand-kyc. No spelling trick is needed, and these are among the most common in practice because they read naturally.

TLD swapping keeps the brand name exactly but uses a different ending: the real company has a .com, the attacker registers the same name under another top-level domain.

Links that hide their destination

Sometimes the visible link is honest but the final destination is not.

  • Display text versus target. In HTML email, the text shown can be anything. Hovering, or long-pressing on mobile, reveals the real target.
  • URL shorteners replace the destination with a short code. Many have preview features that show the destination without visiting it.
  • Open redirects on legitimate sites take a destination as a parameter and forward the visitor there. The link begins with a genuinely trusted domain, which helps it pass filters and human inspection, then bounces to the attacker.
  • Trusted hosting platforms. Phishing pages are often hosted on legitimate cloud storage, form builders or website builders. The domain is real and reputable; the page on it is not.

How an analyst examines a suspicious link

A SOC analyst handed a reported link works through it without clicking on their own workstation.

  1. Defang and parse. Copy the URL as text, often “defanged” with brackets around the dots so it cannot be accidentally clicked. Identify the registrable domain.
  2. Check the domain’s age and registration. WHOIS or registration data often shows a creation date. A domain registered three days ago that imitates a bank is close to conclusive.
  3. Check certificates. Certificate Transparency logs record publicly trusted TLS certificates. Searching them reveals when a certificate was issued for the domain and which other similar names the same actor may have set up. A padlock icon proves only that the connection is encrypted, not that the site is legitimate; free certificates are easy to get.
  4. Check reputation. Look the domain and URL up in threat intelligence and reputation services.
  5. Detonate safely. If needed, open it in an isolated sandbox or a URL analysis service that records redirects, the final page, screenshots and any downloads.
  6. Scope. Search mail and proxy logs for anyone else who received or visited the domain.

Getting ahead of it

Organisations can watch for lookalikes of their own brand before they are used. Tools that generate permutations of a domain (swapped letters, homoglyphs, added words) and check which are registered, combined with monitoring Certificate Transparency logs for new certificates containing the brand name, give early warning. Some companies defensively register the most obvious variants, though there are always more permutations than anyone can buy.

For everyone else, the habit that protects best is the simplest: read the domain from the right, stop at the registrable name, and ask whether that is really who you expected.

Written by the RP Strativa Team.

Leave a Reply

Trending