At the recent Anti-Phishing Working Group meeting in San Francisco, Rod Rasmussen and I published our latest APWG Global Phishing Survey. Phishing is a distinct kind of e-crime, one that’s possible to measure and analyze in depth. Our report is a look at how criminals act and react, and what the implications are for the domain name industry.

This report seeks to understand trends and their significance by quantifying the scope of the global phishing problem, specifically all the phishing attacks detected in the first half of 2013. Here’s some of what we found.

Phishers seek out vulnerable resources. There were at least 72,758 unique phishing attacks worldwide, occurring on 53,685 unique domain names. Most of those domains were hacked—on vulnerable web servers that the phishers broke into. In fact, 27% of all phishing attacks recorded worldwide involved mass break-ins at vulnerable hosting centers. Breaking into such hosting is a high-yield activity, and fits into a larger trend where criminals turn compromised servers at hosting facilities into weapons. We see such servers being utilized for all manner of abuse beyond phishing, ranging from underground proxy networks to large-scale DDoS attacks.

Like other criminals, phishers seek new markets. Phishing is exploding in China, where the expanding middle class is using e-commerce more often. We identified 12,175 domain names that we believe were registered maliciously, by phishers. Of those, at least 8,240 (68%) were registered to phish Chinese targets: services and sites in China that serve a primarily Chinese customer base.

Anti-abuse responses work. Some TLDs have anti-abuse monitoring and response programs, and the data shows that phish in these TLDs are taken down much more quickly, thereby saving a lot of victims. Phishing that uses URL shorteners is also way down, because companies such as BIT.LY and T.CO have implemented good monitoring programs, which has driven phishers away. And a lack of abuse monitoring becomes really noticeable in some TLDs, which suffer high levels of abuse.

Sometimes the data explodes common perceptions. Phishers don’t cyber-squat much. Just 2.3% of all domains that were used for phishing contain a relevant brand name or reasonable variation thereof (often a misspelling). Instead, phishers usually register nonsense strings. Placing brand names or variations thereof in the domain name itself is not a favored tactic since brand owners are proactively scanning Internet zone files for their brand names. Instead, phishers often place brand names in subdomains or subdirectories.

And while people used to worry that IDNs would be used widely to fool people into visiting look-alike domains, the actual occurrence is vanishingly small. In seven years of research, we have found only eight IDNs domains used for homographic attacks, out of the hundreds of millions of domains registered during that time. It’s a reminder that a security vulnerability can be measured by its actual impact.

If you’re a registry operator, a new TLD applicant, a registrar, reseller, or responder, take a few minutes to read the report. It’s a good way to know the enemy, and to protect your company and your users.