Spam Filtering and Email Junk 101

Aug 24
Spam Filtering and Email Junk 101

I’m always getting bombarded with people asking about Spam Filtering and what ‘settings’ have been changed due to the fact that their many ‘homegrown’ mailouts are going into peoples spam filters, so I’m trying this post to demystify and explain how the system all works and how you can do your bit to make sure your emails go into peoples inboxes, not spam/junk folder.

First, you need to understand that most spam filters work using an extremely complex mathematical formula called Bayesian Filtering.

The formula calculates the ‘probability‘ that a message is spam based on the forumlaic equation :

Bayesian Filtering Formula

Bayesian Filtering Formula

Emails are categorised as either spam (or unsolicited commercial email) or ham (normal acceptable emails) – note, I didn’t come up with the spam/ham names, this is how they are referrred. (for those who remember tins of spam of course, you’ll understand the relevance)

So, since an email can either be spam or ham – it’s possibility of being spam is 50/50 – therefore the filter bases on a start value of 50% chance of it being spam.

Based on keyword combinations and density, repetitions of phrases and such like – if the probability score goes over a set value, it is tagged as ‘probably’ being spam and therefore moved to the users junk folder.

This is now where we get ‘one’ of the problems with this system.

EVERY email client has a different tolerance rating!

Some email clients are set at 65%, some 70% some are even set at 80%. The lower the tolerance rating, the more email is likely to be tagged as spam. (Or supposedly, the more ‘secure’ your emails are) Outlook for example (on the PC), you can set your ‘junk’ mail settings as being Low, Medium or High – all this does is adjusts the tolerance of the probability score percentage.

This means of course, that just because it gets caught in your spam folder, it may not in others.

The next problem you have is differences in browser versions. Yes, thats right – Browser Versions!

A user running an older version of outlook and IE6, will be using the IE6 rendering capabilities, therefore if the spam filter detects that the mailout uses things (such as bullet points for example) and IE6′s rendering engine doesn’t support them the way they’ve been written, it will simply add to the spam probability score.

The same for Apple clients, Mail uses Safari as it’s rendering engine, therefore is limited to what it can do by Safari web browser.

Now – we have the biggest curve ball of all time. The later versions of Outlook don’t use IE at all – but they use MS Word

This now means that if you want your emails to reach the maximum of recipients, you have to deal with the lowest common denominator – which for most us is Microsoft Word.

This means the following:

No Text on top of Graphics
You can’t have text on top of big sweeping graphics – because the technique used for this (background imaging) is not supported.

No Bullet Points
You can’t use Bullet points to list items, unless you are happy with the bullets showing. There is no way to hide them.

Text can’t be ‘floated’
You can’t use text which is floated right aligned etc All text must be in tables in a linear format.

This is not exhaustive, but is common with what I’ve seen in the past and the problems that we seem to get caught with.

Checking Keywords is not a problem
A valid question though, could you just change/replace the offending spam words with others to counteract the problem? The problem is that some of the top words and phrases in spam filters (apart from the usual viagra etc) are:

  • Affordable
  • Brand
  • Offer
  • Opportunity
  • Performance
  • Investment
  • Money
  • Trial
  • Bargain

I have spent a significant amount of time discussing mailing lists and mailouts with many clients and I’ve always said mailing communications are crucial to increasing brand awareness of products, services, events and company – so we need to be extremely smart with them.

1) Use the Website as much as possible

Instead of always putting the entire content of a story into a mailing. Consider having an introductory paragraph or shorter article which then invites users to click a link and view the full release online. The advantage is that you are promoting your website (driving more traffic) and also increasing chances of the user seeing something else of interest. Also promote social interaction. Add options to your mailings for ‘send to friend’ etc. This has less of a chance of being reported for spam.

2) Keep Templates Simple

There should be specific types of templates that you use to keep things as simple and clean as possible. I understand that in some cases it may be necessary to send a press release with everything in the body of the email, however this should essentially follow a simple format:


Body Copy


Disclaimer / Unsubscribe Information

Likewise, if you go along the route of having multiple news stories in a single email mailout, then the format should be as follows:


Intro Copy
Read More

Intro Copy
Read More

Intro Copy
Read More


Disclaimer / Unsubscribe Information

Image Blocking is Standard
This shouldn’t come as a surprise, because you all need to do it in your emails. If an email arrives with images embedded, they will not be shown until the user clicks on ‘show images’ or ‘download images’. This means that they will not see anything until they do that.

Research has shown that as many as 30% of people DONT click on the ‘show images’ button. Therefore, you have to make sure that your email is readable, understandable and has some way of the user getting the information they need without showing images.

In addition to that, there are some corporate environments, where HTML formatting of emails is forbidden and servers will automatically strip out all images and formatting and simply deliver the text as a plain alternative.

Remove the Hard Word

Take a look at sites such as Campaign Monitor or Mail Chimp which are excellent applications for sending Mailouts and Newsletters. I have recently moved over our entire group of companies into Campaign Monitor and we maintain the best part of 120,000 emails across 7 projects and about 15 mailing lists using Campaign Monitor and it works great!

  • Share/Bookmark

Leave a Reply

Additional comments powered by BackType

Links of Interest!

Here are some links of sites/people that I find interesting...


This page and most pages on this site should comply with w3c validation where appropriate. If you find it doesn't, please notify me