Does Text to Image Ratio Affect Deliverability?
60/40 ratio: Fact or fiction?
A best practice floating around the emailsphere is that marketers should have a 60/40 text to image ratio in their email file. So why should we email marketers and designers give a hoot about finding the perfect balance with our text and images? Because it’s all about the deliverability, baby!
The idea is that image heavy emails without much text can raise a red flag for SPAM filters. This theory was formulated because spammers sometimes display information in large images instead of text so that the filter programs cannot “read” the content. Because of this, most articles warn against sending image only emails and recommend the 60/40 text to image ratio to avoid email deliverability issues.
This 60/40 assertion ignited the blog topic at hand:
Does your text to image ratio within your email affect deliverability?
We never guesstimate when it comes to best practices for email marketing, so the only way we could find out if a relationship exists between one’s text to image ratio and deliverability was to get to testing with our pre-send SPAM testing tools. Read on to see what goods we discovered!
How we created our test emails.
Before we could start testing the text to image ratio and its effects on deliverability, we first had to find a “perfect” text only email that passed all SPAM filters. That way, when testing we would know that an email got caught in a SPAM filter from a change in the text to image ratio and not another underlying factor. This was a little bit of an uphill battle to appease each and every SPAM filter but we did it!
Now that we had an email that would pass through all SPAM filters, we created email files that ranged in the number of characters and the number of images used.
Before we added images to our files, we made 16 different content-only files that ranged from 168 – 111,190 characters. It wasn’t technically possible to test an email with 0 characters because it would have no content, so our smallest was a bare bones email at 168 characters.
We then cloned each file three times and added images so each email had either no images, one image, two images or three images.
For example, one of our original content-only files consisted of 168 characters. After cloning and creating all the different variations, the four files types existed below:
- 168 character file with 0 images (168 total characters)
- 168 character file with 1 image (267 total characters)
- 168 character file with 2 images (354 total characters)
- 168 character file with 3 images (441 total characters)
These variations existed for all of the 16 original files which equated to 64 files total.
Because our previous research showed us that image size does not affect deliverability, we did not vary the image dimensions in the tests.
Here’s what we found…
By leveraging our SPAM reporting, which tests against 23 of the most popular SPAM filters, we found that if your email has 500 characters or more, content to image ratio does NOT affect deliverability! This finding was contrary to what we’d heard from many marketing resources, so we were surprised to say the least. All of the emails we tested with 500 characters or more got the green light from all the SPAM filters, regardless the number of images added.
It’s critical to note that HTML emails of 500 characters or less are very rare. SPAM filters probably have no problem flagging emails under 500 characters as SPAM because emails structured around this character count usually are SPAM!
A SPAM email often consists of 1-3 sentences with a single link within it. It took me about 5 seconds to search in my junk folder to find a SPAM email structured this way. Check it out below:
So word of the wise y’all, don’t send marketing emails with 500 characters or less. After we uncovered this breaking point, we wanted to look closer and see if text to image ratio for emails UNDER 500 characters or less affected deliverability.
We found that if you are sending an email with 500 characters or less, make sure you have a supporting image within the email. We came to this conclusion because when an email file was comprised of 500 characters or less and did NOT have any images within the email, each email was caught in the 4 SPAM filters below:
- Google Apps
- Outlook 2007
- Outlook 2010
- Outlook 2013
Once one image was added to the mix, the number of SPAM filters that blocked an email with 500 characters or less decreased from 4 filters to 3 filters:
- Outlook 2007
- Outlook 2010
- Outlook 2013
Our next step of testing was to examine if adding more images to files with 500 characters or less would further decrease the number of SPAM filters that failed. Alas, no dice. The number of SPAM filters failed did NOT change when more than one image was added to an email with 500 characters or less. Emails with one, two or three images will still get stuck in Outlook 2007, 2010, and 2013.
It seems the Outlooks turns their noses up at emails with sparse text, whether or not images are in the mix. Moral of the story, don’t send emails with 500 characters or less. If you follow this simple rule, your text to image ratio won’t affect your deliverability!
If you feel you ABSOLUTELY must send an email under 500 characters, make sure you have at least one image in your email. However, adding more images won’t up your chances of passing through all the SPAM filters.
Are your emails reaching the inbox?
You can (and absolutely should) follow best email practices to avoid deliverability issues, but how do you know if that’s enough to ensure your email reaches the inbox? Most email marketers aren’t even aware they are having issues with SPAM filters until a serious problem rears its ugly head, like having your domain end up on a blacklist or your sender reputation taking a nose dive. That’s where pre-send email tools like SPAM testing come in to play.
The only way you can know you’re having deliverability issues is by testing BEFORE you send. Protect your ROI and your bottom line by leveraging our SPAM testing software to ensure your inbox placement.
What issues have you seen regarding text to image ratio and deliverability? Have you done testing on SPAM filters and seen different results? We’d love to for you to comment below regarding the quirks you have encountered in the wild or the workarounds you’ve found along the way.
Author: Kyle Lapaglia
Author: Kyle Lapaglia
25 thoughts on “Does Text to Image Ratio Affect Deliverability?”
Comments are closed.
Interesting stuff, glad you put it all to the test!
The image to text ratio thing has generally been a thing for stand alone filters like Outlook client and Spam Assassin.
The larger free inbox providers like Hotmail Gmail etc. will base their content leniency on reputation and other content factors, eg: if you’re spammy you don’t get to use lots of images.
Spam Assassin still seems to throw a little warning on a single image, although it’s a low score.
eg:
2.5k chars, one image, spam assassin says:
“BODY: HTML: images with 2800-3200 bytes of words”
It’s worth adding that most open/render tracking uses a tracking pixel which is normally invisible, so even if you only add text, you get a single image.
In my experience of Spam Assassin having at least 3 images and not all touching, seems to please it, 2 images not being enough.
If you think about the evolution of the spam filter as spammers would work out how they identified them and altered their tactics it kind of makes sense.
I did a little write up; you’re welcome to the copy as long as you give credit to me & Pure360 😉
http://getintheinbox.com/2010/03/26/email-marketing-deliverability-spammers-vs-the-spam-filter/
Love your work
Andy
Now the funny thing is senior figure at Cloudmark in the UK was touting this story too – but most of us know it’s not true. The text in images was dealth with years ago with fuzzy ocr etc – but I guess they’re still in the 90’s….
Very interesting stuff, but I tend to agree with Andy.
Perhaps there might be some overriding factors which has an affect on your numbers – for example the Sender Reputation of the mailing list in particular?
i.e. If you send to a mailing list with really good reputation, then that might perhaps bring that 500 character down, alternatively, if you have a really poor one then you might need to increase things.
By 500 characters – are you counting that in the html code, or actual displayed text? How have you defined images as 90 odd characters?
Stephen,
I absolutely agree with the points you made. Sender reputation greatly effects your deliverability. Someone with a good sender reputation will most certainly bypass more SPAM filters than someone with a crappy reputation.
It’s better to be safe than sorry, though, when trying to get your email past SPAM filters, so I just wanted to lay out best practices to avoid getting tied up in these filters 🙂
I have to agree with @CaptainInbox on this one… it’s a hold over from Spam Assassin that had a number of different ratio sizes as rule definitions. This was part of the standard rule sets that came out-of-the-box with an SA deployment and not included in one of the many customizable and expanded ‘rules emporium’ rule sets. This is however a great test and good learnings. Thank you for sharing!
Hey Lee,
We are counting the characters in the html code, not the actual displayed text. Our images we tested are around 80-100 characters!
I feel like this has taken a but of turn and I’d like to add a bit more context to both points:
Spam Assassin is stand alone, so a self hosted server or a hosting provider may rely on it as part of it’s decision making.
Spam Assassin has recently gone to a nightly update and has been a bit buggy but I think/(want to)believe they’ve got a handle on it now.
None of the big B2C ISPs reference it directly but have evolved at the same time for the same reasons (except Gmail which was later but has better content parsing technology).
So yes, leniency of content filters at Gmail, Hotmail Yahoo etc. rely heavily on reputation.
If you find you’re getting junked, clean up against Spam Assassin first. It’s the easiest variable to solve. If your rep’s dropped recently, you’ll have less freedom and Spam Assassin can point out where to clean up.
So having Spam Assassin warnings does not mean you’ll be junked everywhere by any means, it simply heightens the likelihood.
B2B will be more influenced than B2C.
+ Gmail and MessageLab do B2B and rep.
Do your own testing and see.
This was a great post Mallory, there’s no replacement for real testing.
My takeaway from this is to build a safe foundation and then widen the circle with creativity and test test test.
Just curious, how are you calculating the defined characters of an image? Say I have an image that is 600px x 100px…
Hi Brian,
We are counting the characters in the html code, so the pixel dimensions of the image are unrelated. Let me know if you have any further questions!
Really, 500 words for a clean email? This will make us reconsider email-marketing strategy!
I do want to clarify that we recommend you use 500 characters or more, not 500 words or more. This method won’t guarantee a clean email though because outliers such as sender reputation still effect whether your email will pass through the filters.
When you refer to including images, are you embedding the images or linking to them on an FTP site?
The approach you take to including images drastically alters the text to image ratio.
Hey Alis,
We used images stored on a server for these tests. We added it to the email using image tags. For example, <img src=”www.emailonacid.com/image01″ />. We have not tested embedded images. Please let me know if you have anymore questions.
Hi. Nice article. Just to say that the text to ratio image doesn’t resume to a spam filter problem, but also to a user engagement issue. Since a lot of webmails and email clients block images by default, if there is too many images among text in your email, this later will be unreadable ! 😉
(and yes, I know you have a Mozify option ^^)
Hi – I have a question for the character placement – would it count if these be just placed in the area alt tags? Just curious if that would have any impact on making it into the inbox.
Thank you!
Monja,
Even if they were placed in the alt tags, it would still count for the number of characters.
Let me know if you have anymore questions!
I am completely agree with thanks for sharing with us.
Mallory-
Is there any update to this?
I got a reader comment from an article I wrote ( http://www.targetmarketingmag.com/?p=31765 ) about your post. This was great information and a well thought out test.
By the way I love Email on Acid. Your product , you info, everything. Can’t believe I missed this post.
Patrick
Hi Patrick,
Thanks so much for the kind words, I really appreciate it! We haven’t done any re-testing on this subject but it would be a good idea to do so in the near future for us to stay up to date 🙂
Have a great weekend!
Mallory-
If and when you do update it, please let me know. Given it’s original date I wouldn’t be surprised if it radically different. Although eMail tend to change slower than anything else.
Patrick
Mallory – I’m curious to know if you’ve had a change to retest this, per Patrick’s suggestion. Thanks! Frank
Hey Frank, I hope you’re well. We still have not run round 2 of these tests but will update this blog when we do. Please let me know if you have any other questions!
Best info I’ve seen so far on this. I’ll try to increase character count and hope it works. Thanks!
Hey Mallory! Anything new on this front? I’ve created an email with MJML and my finished product has 24,417 characters. This is a new template for me, previously I’ve used traditional HTML and had only a couple of thousand characters. Weirdly, for the first time ever, Send Forensics (my spam pre-tester) has been telling me that my text/image ratio is too low. Do you think I can ignore this since my email is way over 500 characters? Not sure what to think about this!