FAIR A/B TESTING FOR EMAIL
See which version wins, for real
Every campaign can go out as several designed versions, not just two subject lines, and each one reaches an even slice of your whole list. No small sample, no early guess. You see exactly which version your audience preferred, counted from real people opening and clicking.
Rotated evenly across your whole audience
Human opens counted separately from automated ones
You pick the winner, not the clock
THE OLD WAY
Most tests crown the fastest opener, not the best email
Here's what a typical test does: it sends two subject lines to a small slice of your list, waits about an hour, and declares a winner based on whichever one got opened first. The rest of your list, the people who actually matter, never get tested at all. They just get whatever won that hour.
The trouble is, the fastest open usually isn't a person at all. Security scanners and inbox previews open mail within seconds of delivery, long before anyone's had their coffee. If your test counts those opens the same as a real reader, you've crowned a winner that no customer actually chose.
And a test that only changes the subject line misses everything else that moves someone to click: the opening line, the offer, the layout. A real test should let whole designs compete, across everyone, and count the people who actually looked.
HOW IT WORKS
Built to run a fair test, every time
Set up your variants once. Six and Two handles the rotation, the counting and the reporting while your campaign sends.
Design your variants
Build two, three or more versions of the same campaign: a different subject line, opening, offer or layout for each one.
Launch to everyone
Variants rotate evenly across your whole audience as the campaign sends, not a small test group first.
Watch results separate
Each variant reports its own opens, human and automated kept apart, and its own clicks, as they come in.
Choose your winner
Nothing gets crowned on early numbers. You compare the real results and decide which version worked.
IN DETAIL
Three ways this keeps the test honest
Every part of a campaign can carry its own variant, every variant reaches everyone, and the results are reported straight.
DESIGNED VARIANTS
More than a different subject line
A campaign built with Six and Two can launch with several fully designed variants: a different subject line, a different opening line, a different offer, even a different layout, all in the same send.
That means you're testing what actually changes someone's mind, not just which four words got clicked in an inbox. Try a discount against a free-shipping offer, or a short opening against a longer story, and see which one your audience responds to.
Subject line, opening, offer or layout, each a variant
Build variants in the same drag-and-drop editor
Every variant ships from the same campaign
FAIR AT FULL SIZE
Tested on everyone, not a slice
Variants rotate evenly across your whole audience as the campaign sends. There's no small sample held back for a quick read and no guesswork applied to the rest of the list. Every subscriber is part of the test.
That also means your send doesn't look like one identical blast landing in thousands of inboxes at once. Different subject lines and layouts going out together read as normal mail to inbox providers, not a single pattern repeated at scale.
Every subscriber sees one of your variants
No held-back sample, no delayed send
Varied mail, not one identical blast
RESULTS YOU CAN TRUST
You pick the winner
Results come back per variant, with human opens reported separately from the opens that come from security scanners and inbox previews. You're comparing people, not software.
Nothing is crowned automatically on an early, noisy number. You look at the real results next to each other, for as long as you want, and choose the version that actually worked.
Per-variant opens, clicks and results
Human opens separated from automated ones
No auto-crowned winner on early numbers
WHAT IT MEANS FOR YOU
Decisions backed by real responses
Fair testing changes what you can trust about every send.
Confidence in every send
You stop guessing which version will land and start seeing it, every single campaign.
Side-by-side results mean you can spot a pattern across sends, not just one lucky subject line.
Real winners, not scanner picks
Automated opens never get to decide. The version that wins is the one real people actually opened and clicked.
Every subscriber counts
No test group held back and no guesswork for the rest of the list. Everyone's response is part of the result.
Mail that reads as normal
Different versions going out together keep your send from looking like one identical blast to inbox providers.
You make the call
Results are there to read, not to auto-apply. You choose the winner on your own timeline.
COMPARED
A typical test versus a fair one
A typical A/B test
- Small sample. Tests a slice of your list, then guesses for the rest.
- Early numbers crown it. Whichever version opens fastest in the first hour wins.
- One opened subject line. Counts whoever opened first, scanners included.
- Subject lines only. Tests a few words and calls it a test.
- Looks like one blast. Every recipient gets identical mail.
Six and Two
- Full audience. Every variant reaches an even share of everyone, so the test is the send.
- You decide. Nothing is crowned automatically on early, noisy numbers.
- Human opens only. Automated opens are reported separately, never counted as a win.
- Full designs. Variants can differ in subject line, opening, offer or layout.
- Looks human. Different versions going out keep your sending varied to inbox providers.
QUESTIONS
What people ask about variant testing
Is this the same as A/B testing?
It's a fairer version of it: every variant goes to an even share of your whole audience, and results count people, not scanners.
How many variants can I test in one campaign?
You can launch several designed variants in the same campaign, each with its own subject line, opening, offer or layout. Build as many versions as the test needs; Six and Two rotates all of them evenly.
Do variants have to be different subject lines?
No. A variant can change the subject line, the opening, the offer or the whole layout. Test whatever's likely to move your audience, not just the first few words they see.
How is this different from testing a small sample first?
There's no sample. Every variant rotates across your entire audience as the campaign sends, so the test is the real send, not a preview of it.
How do you know a human opened it, not a bot?
Every open is classified. Opens that come from security scanners and inbox previews are reported as automated, kept apart from human opens, so your comparison stays honest.
Who decides the winning variant?
You do. Results are reported per variant as they come in, but nothing is crowned automatically on early numbers. You compare the full results and choose.
Does sending several versions hurt deliverability?
No, it helps. Identical mail landing in thousands of inboxes at once reads as one blast. Different versions going out together look like normal mail to inbox providers.
Is variant testing available on every plan?
Yes. Every feature is on every plan, and plans run month to month in US dollars.
WORKS WITH
Pairs well with the rest of your campaign
Variant testing works best alongside the rest of how Six and Two builds and sends.
AI email design
Start each variant from a one-line brief and let it write in your brand's voice, then tweak what you want to test.
See AI email designHonest reporting
See per-variant opens and clicks, with human activity counted apart from automated opens, live from the first send.
See honest reportingSending domains, handled
Every variant sends from your own domain, registered and authenticated for you, so results reflect your mail, not someone else's.
See sending domainsPre-send review
Every variant gets checked, subject line, structure, links and sender details, before any of them go out.
See pre-send review
Let every version compete fairly
Build your variants, send to everyone, and let the real results pick the winner.
Create your account