Orbit web apps
Orbit web apps
Read the text-forward balance of an email before you send it. Runs entirely in your browser.
Parsed in your browser only. Nothing is uploaded, stored or sent anywhere: close the tab and it's gone.
Why the balance matters
An all-image email is the classic way to hide copy from a content scanner, so filters distrust it, and clients that block images show your reader empty boxes. A text-forward build (the rule of thumb is roughly 60% text) reads as a normal message. The percentages here are indicative, not a spam score: deliverability is reputation-driven. Email deliverability has the wider picture.
Paste email HTML to read the text-to-image balance.
The text-to-image balance of an email is one of the few deliverability signals you can see before you send. It won't predict the inbox. Nothing in an HTML string does that, because deliverability is reputation-driven, but it catches the avoidable version of the mistake: a beautiful, entirely-image email that a content filter reads as a message with something to hide.
The reason an all-image email is treated with suspicion is mechanical, not aesthetic. Spam filters scan the text of a message for signals. If the entire message is a single exported image (the copy, the offer, the call to action all baked into a JPEG), there is no text to scan, which is precisely how a spammer hides a message from a content filter. The filter can't tell your polished campaign from that, so it leans cautious.
The rough guideline people cite is roughly 60% text to 40% image, and it is a guideline, not a threshold anyone published as law. What matters is the direction: enough live HTML text that the message reads as a normal email. The checker above weights each content image against the copy to give you an indicative share, and excludes the 1×1 tracking pixels and layout spacers that would otherwise inflate the image count.
A large share of inboxes don't load images until the reader clicks to show them, and some never do without a per-sender allow. For an image-only email, that means the reader's first impression is a stack of empty rectangles with broken-image icons: no headline, no offer, no reason to click.
Alt text is the fallback that shows in that gap, which is why the checker reports alt-text coverage as a first-class number rather than an accessibility footnote. A decorative image can legitimately carry an empty alt, but a content image without one is a silent hole in the message for every reader who has images off, and for every screen reader.
A well-formed marketing email is multipart: an HTML part and a text/plain part, and the receiving client picks. Most ESPs generate that plain-text part automatically from your HTML. If the HTML is all images with almost no live text, the auto-generated plain-text part is nearly empty, and a missing or empty plain-text alternative is itself a spam signal, on top of being a blank message for text-only clients and some assistive tools.
From an HTML paste alone the tool can't see the MIME structure, so it's honest about that: it tells you whether a usable plain-text part is derivable from the copy you have. Rich copy means a plain-text part that carries the message; near-zero text means one that says nothing. If you paste a full multipart source with a real text/plain section, it detects and confirms it instead.
Zero clickable links is worth catching, because a call to action rendered as a flat image isn't a link: there's nothing to click, and nothing to attribute. With Apple Mail Privacy Protection inflating open rates by pre-fetching images, clicks are the signal that actually measures an email programme, and they only exist if the buttons are real links.
At the other end, a very high link count, especially links pointing at many different domains, is one of the oldest spam tells. Legitimate for a content digest, a red flag on a simple promotional send. The checker reports both the total and the unique-domain-free count so you can tell an intentional digest from an accident.
Nothing here is a spam score. Two identical emails can land differently depending on the sending domain's reputation, the recipient's past engagement, and the provider's own model, none of which live in the HTML. Anyone selling you a single number that predicts the inbox is selling you a number.
What the checker does is remove the avoidable mistakes before they ship: the accidental all-image send, the images with no alt text, the empty plain-text part, the call to action that isn't a link. Get those right and the balance stops being the thing holding a good email out of the inbox.
Built into Orbit
Orbit's render and deliverability tooling runs this balance check as part of a wider pre-send gate (dark-mode and client rendering, image-to-text weight, alt-text coverage and auth) against your own email design system. Claude can run the gate on a draft and list exactly what to fix before it ships.
Sign up free