A/B testing, built in

We test two versions of your emails and keep the one that gets replies.

Nobody knows in advance which subject line your buyers answer. The big studies even disagree on whether a question beats a statement. So Pipefire writes both and splits your prospects between them. Once each version has 300 sends and one is ahead by a real margin, every new email goes out as the winner.

⚡ Start testing my emails See pricing

app.pipefire.ai / campaigns

Subject line test roofers, Denver

Version	Subject	Sent	Positive replies	Result
Question	roof calls after storms?	318	9 (2.8%)	Winner
Statement	storm damage on rentals	312	3 (1.0%)	Retired

Over 300 sends each, ahead with 95% confidence. Every new prospect now gets the question.

1We test two subject styles on your own prospects.
2We count positive replies, never opens.
3The winner takes over once the result is real.

Example screen with invented numbers: two subject styles, their sends and positive replies, and the result.

What you get

We test on replies, not opens

Opens are faked by Apple Mail and spam filters. We count interested replies, questions and referrals: the ones that become calls.

We wait for a real result

At least 300 sends per version and 95% confidence before declaring a winner. A lucky week never retires a good email.

We switch for you

When one version wins, every new prospect gets it. You see why, in plain words, on the campaign page.

How it works

We write two versions

The same email, two subject styles: a question about their situation, and a plain statement.

We split your prospects

Each prospect gets one version for the whole sequence, so a reply is never credited to the wrong one.

We count the good replies

Interested, a question, or a referral. "Not interested" and unsubscribes do not count as wins.

We keep the winner

With 300+ sends each and a clear gap, the loser is retired and the winner sends from then on.

What runs on its own, and what you decide

Pipefire does this for you

Writing both versions
Splitting prospects evenly
Counting positive replies
Switching to the winner

You stay in charge of

Approving both versions before anything sends
Editing or pausing either one
Seeing the numbers on the campaign page

When something goes wrong

The two versions do about the same

No winner is declared. Both keep sending, and the page says the difference is too small to call.

One version gets an early lucky streak

Nothing changes until 300 sends each. Early streaks are usually noise.

You prefer the losing version

You can edit it or keep it. The result is a recommendation you can see, not a lock.

An example, with an invented business

Testing subject lines for a roofing campaign

Week 1Sam approves both: "roof calls after storms?" and "storm damage on rentals".
Week 2Each version has about 150 sends. The page says it is too early to call.
Week 4Over 300 sends each. The question gets 2.8% positive replies, the statement 1.1%.
Week 4The question version wins and every new prospect gets it. Sam is told why.

Doing it yourself vs Pipefire

Doing it yourself	Pipefire⚡ Done for you
Guess which subject line works	Two versions tested on your own prospects
Judge by open rates that are mostly fake	Judged on positive replies only
Call a winner after 20 sends	300 sends each and 95% confidence first
Remember to switch versions by hand	The winner takes over by itself

Questions

What exactly is tested?

The subject line style: a question about their situation against a plain statement. Every version still follows the same writing rules.

Why not test on open rates?

Apple Mail and many spam filters open emails automatically, so open rates mostly measure machines. Replies are people.

How long until there is a winner?

When each version has at least 300 sends and one is ahead with 95% confidence. On a full Standard setup that is usually a few weeks. Until then, the page says there is no winner yet, and why.

Can I stop a test?

Yes. You can edit or pause either version at any time.

How it fits with the rest of Pipefire

← Before thisEmail writingYou answer a few questions. We write cold emails for your prospects, ready for you to approve.Next →Bounce and spam protectionWe verify every address before sending, and pause on our own if bounces pass 3% or complaints pass 0.1%.

Everything else Pipefire does for you

Your agent team

Eight agents, each with one job: find, check, set up, warm, send, guard, reply and supervise.

See how it works

Domains and mailboxes

We buy 10 sending domains, set up 20 mailboxes and write every DNS record. Your real domain never sends cold email.

See how it works

Warm-up

We warm every mailbox for 14 days and check it daily before it sends a single cold email.

See how it works

Prospect finding

Pick a trade and an area. We find the businesses, the decision-maker and a verified email address, fresh, never from a bought list.

See how it works

Email writing

You answer a few questions. We write cold emails for your prospects, ready for you to approve.

See how it works

Bounce and spam protection

We verify every address before sending, and pause on our own if bounces pass 3% or complaints pass 0.1%.

See how it works

Reply inbox

We sort every reply, put interested prospects and questions at the top, and draft an answer for you to send.

See how it works

24/7 monitoring

We watch every mailbox and domain around the clock, and deal with problems before you notice them.

See how it works

Your real domain never touches cold email.

Separate domains, warmed and watched for you, so the domain your clients email stays clean.

⚡ Start testing my emails See a worked example