Newsletter image to text ratio: what 12,000 sends show

What "image to text ratio" actually measures

There are three things people call newsletter image to text ratio and only one of them maps to deliverability scoring.

The first is byte ratio: the size of all embedded image bytes divided by the total email payload. This is what most spam filters look at when they decide whether your message looks like an image with little body content. SpamAssassin scores HTML_IMAGE_RATIO_02 through _08 using this measure. Gmail and Outlook check the same thing in a way they do not publish but that we can reverse engineer from observed scoring shifts.

The second is pixel area ratio: how much of the rendered email is image versus text on screen. This is what designers usually mean when they say a newsletter is image heavy. They are eyeballing the layout, not opening the source.

The third is DOM node ratio: how many <img> tags exist relative to total HTML nodes. This is the weakest signal. A single 600 KB hero image can be one <img> tag and still get you classified as image-heavy by every filter that matters.

When we say image to text ratio in the rest of this article, we mean byte ratio. It is the one filters score in practice and the one we can measure consistently across the corpus.

The 60/40 rule has no data behind it

Every guide on page one of Google recites the 60/40 newsletter image to text ratio rule, or its cousin the 80/20. Trace those numbers back and you hit blog posts from 2014 quoting other blog posts from 2013 quoting an unnamed "industry best practice". There is no underlying study.

We pulled byte ratios across 12,000 newsletter sends in the Newsletrix corpus, captured between January and April 2026. The actual distribution: median 41% image bytes, P25 18%, P75 67%. The shape is bimodal, not bell. There is a text-first cluster around 15 to 25% (newsletter writers using markdown-style templates) and an image-heavy cluster around 60 to 75% (ecommerce campaigns and creator product launches).

The mean is misleading on this data. The median tells you that the typical newsletter has 41% image bytes, well above the supposed 40% ceiling but well below the supposed 80% ceiling. The bimodality tells you the rule cannot work universally because operators are clearly running two different playbooks.

Image ratio by ESP: Mailchimp vs Klaviyo vs beehiiv vs SendGrid vs Brevo

Where you write determines what you ship. We segmented the corpus by detected sending platform using the ESP detector and the per-platform medians fall out clearly.

Klaviyo lands at 68% median (P25 51%, P75 79%). The ecommerce default templates ship with product grids and lifestyle heroes. The starter Klaviyo template that ships with the platform is already at 60% image bytes empty.

Mailchimp sits at 52% (P25 35%, P75 68%). The drag-and-drop builder favors image blocks. Operators who hand-edit the HTML come in lower.

Beehiiv hits 22% (P25 12%, P75 38%). The text-first editor and creator-oriented templates push the lowest ratios in our corpus. The default footer image accounts for the bulk of total image bytes on most beehiiv sends.

SendGrid medians at 38% (P25 22%, P75 54%). The number is heavily skewed by transactional templates that come through SendGrid alongside marketing emails. Filter to marketing only and the median climbs to 51%.

Brevo medians at 46% (P25 28%, P75 61%). Library templates sit close to the corpus median.

The takeaway: an image-heavy Klaviyo email is normal. The same ratio on a beehiiv send is two standard deviations from the platform norm and worth a second look. ESP defaults encode an editorial stance. Pick the one whose default sits closest to where you want to land.

Ratio vs spam score: 8 paired examples from our corpus

Spam filter scoring is more sensitive at the extremes than in the middle. To show this, we paired byte ratios from 8 corpus newsletters (March 2026 sends, sender identity removed by vertical) with the output of the Newsletrix spam-score checker.

Vertical	ESP	Image bytes	Unique text words	Spam score (0-10)
Ecommerce	Klaviyo	78%	42	6.4 (risk)
Ecommerce	Klaviyo	71%	138	3.1 (low)
Creator	beehiiv	18%	720	1.0 (low)
Media	Mailchimp	54%	410	1.8 (low)
SaaS	Customer.io	32%	280	1.4 (low)
Creator	ConvertKit	25%	580	0.9 (low)
Ecommerce	Klaviyo	82%	28	7.9 (high)
DTC supplements	Klaviyo	73%	380	2.8 (low)

The two outliers at 6.4 and 7.9 share one trait: image bytes above 70% combined with under 50 words of unique text. The 71% and 73% rows are fine because they pair high images with substantive copy. Ratio alone does not predict risk. Ratio combined with low text volume does.

This is the threshold worth memorizing. 70% image bytes and under 50 words is the danger zone. Either lower the image weight or raise the text. Either side works. Spam filter mechanics goes deeper on what else moves the score, including link density and header alignment.

Why alt-text coverage moves the threshold

Alt-text is read by spam filters as additional text content. When alt-text is present and descriptive, the effective text-to-byte balance rises even if the visible body stays short. We have a separate alt-text best practices piece on how to write it. The corpus pattern is clear: newsletters with 70% image bytes and complete alt-text coverage score lower in our spam checker than newsletters with 50% image bytes and zero alt-text on the heroes.

The honest version of the rule: image weight without alt-text is the actual signal. A 60% image email with empty alt attributes is more deliverability risk than a 75% image email where every <img> has a sentence-level description. We see this regularly when we audit client sends. The alt-text auditor flags any image where the attribute is missing or set to a filename like image1.jpg.

Vertical-specific ceilings

One number cannot serve every vertical because reader expectation differs.

Ecommerce subscribers expect a product grid. Klaviyo's 68% median is normal for these readers because that is what they click through to buy. Force ecommerce below 50% and you ship a wall of text that nobody reads. CTR drops, list health drops, and you damage the metric you were trying to protect.

SaaS subscribers expect prose. A 60% image SaaS update reads like marketing fluff. Median SaaS sends in our corpus sit around 28% image bytes and the high-performing ones go lower still.

Media and editorial newsletters sit in the middle, close to the corpus median. Creator newsletters skew text-heavy. Anything above 35% image bytes from a writer-led brand reads as out of character to the audience that subscribed for the prose.

The right rule is industry-specific. Use the per-vertical baselines above, then move from there based on what your readers respond to. Our readability sweet-spot piece covers the matching text-side decisions, and the Newsletrix vs Litmus comparison goes through what each platform measures alongside ratio.

A 3-step audit for your next send

Run this on the next send before it goes out.

First, measure your byte ratio. Total image payload divided by total HTML plus image payload. Most ESP previews tell you the total send size; the image proportion is the part you have to compute yourself.

Second, run the email through the spam-score checker. The score plus the breakdown will tell you whether image ratio is contributing to risk or whether the issue is somewhere else (link density, header inconsistency, body word count, suspicious phrasing). If the image weight is high and the body is thin, the checker will surface both numbers and the spam-score outcome.

Third, audit alt-text coverage. Every <img> tag needs a description that adds reading value. If your team writes generic alt-text and you want help drafting better lines, the AI prompt generator has templates you can feed your image list into.

If you are above 70% image bytes and under 50 words of unique copy, do not send. Add copy or trim images. Pick one.

Frequently asked questions

What is a safe newsletter image to text ratio in 2026?

There is no single safe ratio. The Newsletrix corpus median is 41% image bytes, but Klaviyo ecommerce sends sit naturally near 68% and beehiiv creator sends near 22%. The real risk threshold is image bytes above 70% combined with under 50 words of unique text. Above that line, deliverability degrades regardless of ESP.

Does Gmail penalize image-heavy emails?

Gmail does not publish the exact rule. In our spam-score testing, emails with image bytes above 70% and minimal body text get sorted into Promotions or Spam at noticeably higher rates than balanced emails. Alt-text coverage softens the effect because filters score alt attributes as additional text content.

How do I measure newsletter image to text ratio?

Take the total bytes of all embedded images divided by the total email payload (HTML plus images). Most ESPs report total send size but not the breakdown. The Newsletrix spam-score checker exposes this number for any send you paste in, along with the deliverability score implication.

Why do Klaviyo emails have more images than beehiiv?

Klaviyo is built for ecommerce product marketing. Its default templates ship with hero shots and product grids, so image bytes are high before an operator adds anything. Beehiiv is built for newsletter writers, so its text-first templates and creator-oriented blocks default to lower image weight. The platforms encode different content models.

Can alt text compensate for a high image ratio?

Partly. Alt-text reads as additional text content for spam filters, so a 70% image email with complete alt-text coverage often scores better than a 55% image email with no alt-text. It does not fully compensate at the extreme (above 75% image bytes with under 30 words of body), but it widens the safe zone meaningfully.

Related reading