PerfectMail is Intelligent

PerfectMail is proof that it is possible to develop a fast, consistent, reliable and trustworthy antispam solution. To achieve these objectives, we had to take a new and unique approach.

Before we look at what makes Perfectmail so special, lets look at what is wrong with the traditional approach to spam filtering.

What's Wrong With the Traditional Antispam Approach?

The current crop of antispam tools are simply not good enough. If they were, the Spam problem would be solved and we could all move on to something more interesting.

In our view, the main reasons competing antispam tools are not sufficiently effective are two fold; they do not retain historical information and, even more fundamentally, they take the wrong approach.

As a result of these deficiencies, traditional antispam tools suffer from:

• Analysis Paralysis. Without the ability to perform historical checks, the tool has no choice but to perform an exhaustive series of tests on an e-mail, sometimes numbering 200 or more tests per e-mail. The result is a filter that consumes excessive CPU resources looking the slightest flaw.

  Spammers have long since learned to get around these tests by stacking their e-mail with content designed to confuse scanners and allow their e-mail through.

• Inconsistent Results. Because historical information in unavailable, the filter is unable to make consistent decisions regarding e-mails sent from your mail peers. The result is reduced accuracy and the risk of lost or delayed e-mails.

• Quarantines. Quarantines are an e-mail users nightmare. Simply put, e-mail filters that cannot decide what to do with an e-mail park that e-mail in a quarantine. Then, users must periodically visit, vet and unquarantine e-mails. This wastes user time and (in our opinion) is a strong indicator of the lack of effectiveness of the e-mail filter.

• Antispam Signatures. Many Antispam solutions use the same scanning techniques as antivirus software. It should come as no suprise that most of the vendors who take this approach are in the antivirus business!

  The simple fact is that Antivirus scanning techniques do not work for Antispam. Unlike virus authors (who are limited because their code must be executable by the target computer), spammers have the freedom to change their content when and as they see fit. They can and do introduce randomness into their content to defeat antivirus like spam scanners. Spammers have proven to be very nimble - they can and do change their techniques faster than antispam companies can update their software and signatures.

PerfectMail's Answer To These Problems

PerfectMail addresses all of the above deficiencies. Here is how PerfectMail sets a new standard in antispam filtering:

• Effective Analysis
  Perfectmail includes a suite of over 80 tests. Most of these tests are performed against every e-mail. If a consensus is not achieved, PerfectMail applies additional tests that are more computationally expensive. That way, obviously legitimate (or unwanted) e-mail is handled quickly and effectively.

  PerfectMail is written in C, the fastest executing high level language avalable. So, we can run more tests in less time with lower overhead. And, when PerfectMail decides to run additional tests, you don't pay a big performance penalty.

• Consistent Results
  PerfectMail consults its historical databases before reaching a final decision. E-Mail peers with good e-mail history are shown leniency when PerfectMail is in doubt. Likewise, someone with a history of sending unwanted mail has their e-mail subjected to much more scrutiny and is not given the same consideration.

• No Quarantines
  Quarantines simply don't work. PerfectMail handles all e-mail only once, taking the burden of emptying quarantines from your staff. Furthermore, PerfectMail only has only 3 possible dispositions when dealing with an e-mail:
  1. Accept. The e-mail is passed through to the recipient.
  2. Tag. If, after performing its full range of tests and then consulting its historical databases, PerfectMail is still unable to determine the status of an e-mail, it tags the E-Mail (either subject line or mail header) with a keyword to indicate that the e-mail might be spam. The e-mail is forwarded (with this warning) to the intended recipient.

     Under normal circumstances, PerfectMail tags no more than 0.5 percent of all inbound e-mail. Tag percentages will vary with traffic.

  3. Reject. The e-mail is refused. PerfectMail sends back an SMTP reject code indicating to the sender that the e-mail is unwanted. PerfectMail reserves this treatment for obviously unwanted e-mail such as; viruses & worms, heavily obfuscated e-mail and e-mail where the senders attributes cannot be reliably determined.

• No Spam Signatures
  PerfectMail doesn't use signatures so there is no signature file to update and there are no opportunities for spammers to sneak their content past an un-updated filter. PerfectMail achieves a very high level of scanning accuracy without the need and expense of antivirus-like signature files.

Intuitive, Effective Decision Making

The secret to PerfectMails success is to perform the right tests on the right e-mails at the right time! These tests include traditional tests for spam, new next-generation tests to significantly improve accuracy, and a suite of historical tests designed to ensure the continued delivery of all e-mail from your e-mail peers.

PerfectMail's historical capabilities are unique. Historical checks are not universal in spam filtering but they are the standard of practice in financial environments. For example:

If you were going to grant a customer credit, wouldn't you first insist on a credit check to ensure your customer was credit-worthy?

The assumption is simply that past-history is a solid predictor of future performance. PerfectMail takes the same approach. Past e-mail activity is reviewed and has a major impact on the overall rating of any e-mail under consideration.

PerfectMail then records the current e-mails score (in real time) so that its values can be considered the next time you receive an e-mail from this sender.

Finally, PerfectMail uses techniques designed to discourage spammers. In our own tests, we've seen the volume of unwanted bulk e-mail drop significantly (25+%) once we enabled these techniques.

We'd love to tell you how we do this, but once the word got out, spammers would adapt and we would all end up receiving a lot more spam. (Hint: no, it is not honey-pot e-mail accounts.)