Spammers vs. the spam filter

Andy Thorpe Andy Thorpe heads up Pure’s Customer Accounts Department which proactively and reactively supports and consults Pure’s ever growing customer base. Andy has earned the nickname of ‘Captain Inbox’ as there is very little he doesn’t know about how to get email delivered; from creative tricks, email content and the nature of spam filters all the way to ISP reputation.

Spammers vs. the spam filter

Since email was first in general use spam filters and spammers have been battling it out to gain control over our inboxes.

Spammers, for the uninitiated, send out mass, unsolicited mailings which flood inboxes and, if sent in their thousands or hundred of thousands, cause a slow down in data flow for ISPs.

In an attempt to overcome this, spam filters were put in place to at least restrict if not prevent spam emails from arriving in users’ inboxes.

So who’s winning – the spammers or the filters? Let’s look at the story so far.

The warm-up round

Initially, filters were able to detect if emails were spam by crawling through their content and blocking anything regarded as unsolicited.

As filters can’t read the text hidden in images, spammers hit back by concealing text content in an image to get past the filter.

Spam filters became savvy to this and stopped letting through image-only emails.

Round two

In retaliation, spammers included some text content that was vaguely coherent or even Shakespearean. Not to be defeated, filters upped the pace and blocked emails that contained just one image and some text. Emails that were allowedthrough, then had to contain plain text and non-touching, multiple images with complex HTML.

Protecting the innocent

Legitimate mailers, not wanting to get caught in the crossfire, began designing their emails with a heavier ratio of text to images covering the area of the page.

Spam filters calculated the area of an image using its height and width. For text they stripped out HTML and left text-only content.

Assuming medium sized text is used, the filter calculated the number of characters and then calculated the ratio between the amount of text to image covering the page. If there was a higher image ratio to text, the email was more likely to be blocked.


And then, before accepting emails into the inbox, email clients chose to automatically disable images altogether – putting the power into the hand of the recipient and allowing them to choose whether to download images in an email.

The moral of the story

Where you have normal text in an image, make it actual text and always include a full plain text version of your email – it may be a chore but spammers won’t bother doing it, so you’ll set yourself apart.

And if you’re really conscientious, include a link to view your HTML email in a web browser – a good email reporting system can still track which parts of your emails are clicked on.