Spintax and the safe-to-send score

Spintax writes alternatives into a subject or body. Each lead gets one combination, picked from the campaign, the lead and the step, so the preview, a test email, a retry, threaded follow-ups and the real send all read the same.

Syntax

• {Hi|Hello|Hey}: one option per lead.
• {Hi|Hello {there|again}}: groups can be nested up to 5 levels deep.
• {Hi::3|Hello::1}: weights. Two colons and a whole number from 1 to 100 at the very end of an option; Hi is picked three times as often as Hello. Options without a weight count as 1. (Two colons so times like 10:30 stay text.)
• {|really }: an empty option, so sometimes nothing is added.
• [[First sentence.|Second sentence.|Third sentence.]]: a shuffle block. Every item appears once, in a different order per lead, joined by a space. Items are trimmed.
• [[sep=", " last=" and "|design|development|support]]: a separator between items and an optional different last one, giving “support, design and development”. Put the values in quotes; \n is a line break, so [[sep="\n"|- one|- two|- three]] shuffles lines.
• {{first_name|there}} is a variable, never spintax: its |fallback isn’t an option. Variables, conditions ({% if %}) and spintax work together; conditions are resolved first.
• \{, \}, \|, \[ and \] give literal characters. A [[…]] with no | stays as written.

Mistakes the editor catches

• Unclosed or unmatched braces, empty groups like {} or {|}, empty shuffle items, weights outside 1–100, nesting deeper than 5 and shuffle blocks with more than 10 items. These show under the email as you type and stop saving and launching.
• Warnings only: a group with one option, a shuffle block with more than 6 items (720+ orders to read), and over a billion combinations.

Preview variations

Open “Preview variations” under an email to see the exact number of combinations, a sample of rendered copies (show more, or shuffle for a different sample), each with its own spam check and score, the average similarity between copies (overlap of three-word phrases; lower means more distinct copies), and the spin groups that change the email most. With a saved campaign and a lead list, “Show for 5 real leads” renders it for real leads through the same preview the send uses.

The safe-to-send score

Every email gets a 0–100 score: 80 and up is green, 60–79 amber, under 60 red. It’s an estimate based on known spam-filter signals; it can’t guarantee inbox placement. A low score warns but never blocks launch; only broken syntax, leftover placeholders and a missing opt-out or postal address block.

• Points: spam-trigger words 18, length 12 (subject 4, body 8; about 50–150 words is ideal for a first email), personalisation 10, variation 10, links 8 (0–1 is best), link quality 8 (shorteners, raw IPs, link text that shows a different site, http, click tracking without your own tracking domain, broken links once you run the link check), capitals and punctuation 8, images 6, opt-out and postal address 6, placeholders and unedited template text 6, plain text vs rich 4, first-email rules 4 (no links or images in the first touch, no fake “Re:”).
• Factors that don’t apply (first-email rules on a follow-up) are left out and the rest scaled to 100. Spam words and placeholders are checked across every option and condition branch; the other factors are scored on each sampled variation, and the email’s score is the lowest one. The breakdown shows every factor’s points and how to fix it.
• Setup is a separate 0–100 score for the mailboxes in rotation: SPF 25, DKIM 25, DMARC 20 and warmup 30 (full marks after 14 days with health 80+). It’s shown next to the content score, not mixed in, because you fix it on the Mailboxes page rather than in the copy.

The review step shows the lowest email score and every email’s score before you launch. The free spam checker on our website gives the same content score and variation preview, in your browser, without saving anything.

Trigger words, by category

The spam check highlights every trigger word in your text, in five categories, each with an icon and a label: Urgency (false deadlines), Shady (phishing, scam and malware wording), Overpromise (claims too good to be true), Money (prices, free, cash) and Unnatural (salesy or robotic phrasing). Hover over a highlight, or Tab to it, to see why it’s risky and two or three ways to say it instead. The counts per category feed the spam-words part of the score.