Spintax for Cold Email: What It Is, How Deep to Spin, and What Never to Touch

Spintax is a variant syntax, written as {option one|option two|option three}, that makes each sent copy of a cold email unique. The sending tool picks one option from each group per send, so no two recipients get an identical body. It matters at volume because mailbox filters fingerprint identical bodies arriving across many inboxes, and because past roughly 2,000 cumulative sends of one variant the pattern goes stale and reply rates decay.

We spintax every campaign we run for clients, under a strict internal rulebook, because bad spintax is worse than none: a spun email that renders an ungrammatical sentence, or drifts a number, goes out to a real prospect under the client's name. This post covers how deep to spin each part of the email, what must never be touched, and the QA step that catches broken banks before they ship.

What is spintax in cold email?

Spintax is nested variant syntax embedded directly in the email body. Curly braces mark a spin group, and pipes separate the options: {Quick idea|Got an idea|Had a thought}. At send time the platform picks one option from each group, independently, per recipient. A sentence with three spin groups of three options each can render 27 different ways, and a full email with spin groups on most lines can render thousands of distinct bodies from one template.

The crucial property: every rendered version says exactly the same thing. Spintax varies the wording, never the message. You are rephrasing one idea the dozen ways a human might naturally phrase it, not writing new ideas. That distinction is the entire craft.

A worked example

Here is one sentence with three spin groups:

{Put together|Came up with|Been working on} {a few thoughts|some ideas|a couple of angles} on how {COMPANY} could {reach more|start more conversations with|get in front of more} of the right buyers.

Rendered outputs from that single line:

• Put together a few thoughts on how Acme could reach more of the right buyers.
• Came up with some ideas on how Acme could start more conversations with the right buyers.
• Been working on a couple of angles on how Acme could get in front of more of the right buyers.
• Put together some ideas on how Acme could get in front of more of the right buyers.

Notice that every combination reads as a natural sentence a real person would send, every option fits its slot grammatically, and the company merge tag is untouched. That is the standard: if any single combination reads awkwardly, the bank is broken and must be fixed before launch. Note the braces too: most platforms use different brace conventions for spintax groups and merge tags, and mixing them up sends a literal "{FIRST_NAME}" to a prospect.

Why does spintax matter for deliverability?

Because filters fingerprint bodies. When hundreds of near-identical emails arrive across many inboxes at the same provider, that is the signature of bulk mail, and no clean wording hides it. An unspun template at volume tells every filter it touches: this is a blast. Spintax breaks the fingerprint. Each send looks slightly different, the way individually written emails do, so the pattern-match across the list fails.

The second reason is staleness. In our campaigns, once a single variant crosses roughly 2,000 cumulative sends, the pattern has been seen enough times that it degrades: the wording is stale to the ecosystem and reply rates decay. Spintax extends a winning variant's life by keeping its sends varied, and deepening the spin at that threshold extends it further. This is one specific case of a broader phenomenon we cover in why cold email campaigns decay: everything in outbound wears out with exposure, and variation is the counter.

How deep should you spin each part of the email?

Spin density should vary by section, because different lines carry different amounts of meaning. Housekeeping lines carry almost none, so they tolerate huge variation. Fact lines carry everything, so they get no spin at all. Here is the density map we apply to every campaign:

• CTA line: spin heavily, 10 to 30+ options. CTAs are short and interchangeable ("Mind if I share?", "Want me to run them by you?", "Open to hearing them?"), so a big bank costs nothing and varies the most-repeated line in the campaign. A CTA with only a handful of options is under-spun.
• Opt-out line: spin heavily from a standing bank. Every email carries a reply-based opt-out, and since the line appears in every send, it needs the biggest bank of all. Ours runs to dozens of phrasings of the same courtesy.
• Body: spin freely, multiple spin points per line. Openers, connectors, verb phrases, soft qualifiers, and outcome phrasings can all vary, at word, phrase, and clause level for chunks up to about 5 to 7 words. The one hard limit: no option may change the meaning.
• Fact-carrying sentences: zero spin. See the never-spin list below.

When to deepen the spin

At launch, light body spin is fine. The CTA and opt-out banks are always in, but a new variant at low volume does not need deep body spin, and your effort is better spent on the offer and the list. The investment case changes as sends accumulate:

What should you spin, and what should you never spin?

The core test for any phrase: if I swap this for a synonym, does the meaning, the proof, or the specificity change at all? No means safe to spin. Yes means protect it. When in doubt, don't spin it.

Always safe to spin

• Openers and framing: "Got | Put together | Came up with", "some ideas | a few thoughts".
• Connective phrasing: "quick question | figured I'd reach out | the reason I'm asking is".
• Soft qualifiers: "useful | helpful | valuable | worthwhile".
• Outcome phrasings: "reach more | get in front of more | start more conversations with", provided each option stays grammatical in the slot.
• Sign-offs: "Thanks | Cheers | Best | Speak soon".
• CTAs and opt-out lines: the biggest banks in the email.

Never spin

• Merge tags. Never alter the tag or anything inside it.
• Brand, product, and tool names. Exact spelling and capitalisation, every render.
• Numbers and metrics. A number is a fact, not a phrase. "Around 3 million" and "~3m" are acceptable wordings of the same value; a different value is never acceptable. Metric drift is the single most damaging spintax error, so guard against it harder than anything else.
• Named case studies and results. The name and the result stay locked together as a unit.
• The unique mechanism. The one specific idea that makes the offer distinctive, a guarantee, a pricing model, a proprietary method, must survive intact in every render. Spin the wording around it, never the mechanism itself.
• Literal reply words. If the copy says reply "yes", the word "yes" is protected in every variant, because it is the word the prospect is told to type.

This never-spin list is also why spintax and A/B testing are different tools that must not blur into each other. Spintax varies the wording of one message; an A/B variant tests a genuinely different message, a new angle, offer, or CTA. If your spin options drift the meaning, you are accidentally running an uncontrolled test inside a single variant, and your results become unreadable. Keep the meaning locked in spin, and test message changes properly, as we lay out in our cold email A/B testing guide.

How do you QA spintax before launch?

Render samples and read them as a prospect would. Pick several random combinations from the spin blocks and read each one end to end; every combination must read as a natural, grammatical, sensible email. Most teams skip this step, and it is the one that catches everything, because spintax errors are combination errors: each option looks fine alone and breaks only next to a particular neighbour.

The specific checks we run on every spun email:

1. Mutual substitutability. Every option in a block must read cleanly in that exact slot: no tense mismatches, no grammar breaks, no awkward double-ups with the surrounding words.
2. Even option lengths. Don't pair a 3-word option with a 15-word one in the same block. Uneven lengths break the rhythm and make renders look inconsistent.
3. Register consistency. Every option must match the tone of the original, casual, formal, punchy, whatever it is. One formal option in a casual email sticks out in exactly the renders you never proofread.
4. No metric drift. Check every number in every sampled render against the source. The value never changes.
5. Protected elements intact. Merge tags, names, mechanism, reply words: present and untouched in every sample.

We enforce this because we learned it the hard way. A formatting convention once drifted between our spintax rules and our sending platform's merge-tag syntax, the two use different brace styles, and the mismatch leaked into live copy before we caught it. Nothing teaches you to render samples like discovering that a rule inconsistency shipped to real inboxes. The render-and-read check is now a blocking step in our pre-launch QA, alongside the spam scan and list checks, and no campaign goes out without it.

FAQ

What is spintax in cold email?

Spintax is a syntax for embedding variants inside copy, written as {option one|option two|option three}. Each time the email sends, the tool picks one option per group, so every recipient gets a slightly different body. The message stays the same, the wording varies, and filters can't fingerprint an identical template across the list.

Does spintax improve deliverability?

Yes, at volume. Mailbox filters fingerprint identical or near-identical bodies arriving across many inboxes, which is exactly what an unspun template looks like. Spintax breaks that pattern so each send reads as an individually written email. It doesn't fix a burned domain, but it removes one of the clearest bulk-mail signals.

How deep should spintax be?

It varies by section. Housekeeping lines get the biggest banks: the CTA line gets 10 to 30+ options and the opt-out line uses a full standing bank. The body gets multiple spin points per line at word, phrase, and clause level. Fact-carrying sentences get zero spin. At launch, light body spin is fine.

What should you never spin?

Merge tags, brand and product names, tool names, numbers and metrics, named case studies with their results, the offer's unique mechanism, and any literal reply word the prospect is told to type. If swapping a phrase would change a fact, weaken proof, or lose the offer's specificity, protect it.

When should you deepen spintax?

Once a variant crosses roughly 2,000 cumulative sends. Past that point the pattern goes stale and reply rates decay, so deepen the body spin with more spin points and bigger banks, and add personalisation tokens only where they're genuinely relevant. If you merged near-identical variants, judge the threshold on their pooled send total.

How do you QA spintax?

Render sample combinations and read them as a prospect would. Every option must fit its slot grammatically, options in a block should be roughly even in length, the register must match the original, and no rendered version may change a fact. If any random combination reads awkwardly, fix the bank before launch.