Email campaigns & A/B testing

Campaigns that earn their place in the inbox.

Write with a rich editor, send to a permission-based audience, and measure signed opens and clicks. Run a simple split test or let Transmitly pick the winning variant from a measured sample and send the rest automatically — on a marketing stream that cannot drain transactional capacity.

Included on every plan · no separate campaign product

The workflow

From draft to a result you can defend.

A campaign is snapshotted, rendered per recipient and tracked as an individual message. There is no separate “campaign send” that hides what happened to each address.

  1. 01

    Audience and permission

    Contacts carry marketing unsubscribe state. Suppressed and unsubscribed recipients are excluded at send time, not discovered after the fact.

  2. 02

    Draft and preview

    Use the visual editor or HTML view. Render a real test copy to yourself, or send variant A and B separately.

  3. 03

    Send or schedule

    The audience is snapshotted into a fixed recipient list. The worker creates one marketing message per recipient with its own status and idempotency key.

  4. 04

    Measure and decide

    Opens and clicks are recorded with signed, per-recipient tokens. Variant analytics stay separate, including the test sample used for a winner decision.

POST /v1/broadcasts
curl https://transmitly.io/v1/broadcasts \
  -H "Authorization: Bearer $TRANSMITLY_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "November product update",
    "audience_id": "aud_…",
    "from": "[email protected]",
    "subject": "What changed in November",
    "html": "<p>Hi {{name}},</p>",
    "track_opens": true,
    "track_clicks": true,
    "variant_b_subject": "A shorter November update",
    "winner_enabled": true,
    "winner_metric": "open",
    "winner_wait_minutes": 240,
    "winner_test_percent": 20
  }'

POST /v1/broadcasts/brd_…/send

Campaign tooling

Small surface area, real controls.

Everything needed to run a campaign, without pretending a single score tells you whether the inbox accepted it.

Rich editor + HTML

Headings, lists, links and formatting in a visual editor, with a source view for exact HTML.

Audiences, segments and topics

Target a saved segment, require double opt-in, and let contacts opt out of individual topics. Unsubscribed and suppressed contacts are excluded before sending.

Signed open/click tracking

Per-recipient HMAC tokens. The click target cannot be retargeted, and mailto, tel and fragment links are left alone.

A/B split tests

Override the subject, body or both. Assign variants at snapshot time and compare per-variant results.

Automation journeys

Enroll contacts on audience join or a tracked link click, run durable wait/send steps with conditions, branch into other journeys, and check consent before every send.

Automatic winner sends

Test a sample, wait, then send the better unique open or click rate to the remainder. Ties go to A.

Deliverability context

SPF/DKIM/DMARC audits, BIMI asset checks, DMARC reports, DNSBL state and six-hour DNS change alerts.

Automatic winner sends

Test a sample. Send the winner. Skip the guesswork.

A winner test holds most of the audience while a sample receives A or B. When the window closes, the variant with the higher unique open or click rate among delivered test recipients is sent to the remainder.

Choose the metric

Open rate is faster but noisier because privacy proxies request pixels. Click rate is slower but usually more intentional. Pick one before you send.

Set the sample and window

A five to fifty percent sample, split evenly between A and B, with a window from one minute to 24 hours. Small samples produce noisy winners, so give the test enough recipients to mean something.

Read the decision

The winner, timestamp, metric and both sample rates are stored on the broadcast and emitted as broadcast.winner_selected. Analytics keep the test sample separate from final totals.

No statistical significance theatre. If the sample is thin, treat the result as a hint, not a verdict.

Test sample

automatic winner
VariantDeliveredUnique opensRate
A98014214.5%
B97618318.8%

Illustrative sample data. Variant B wins and is sent to the remaining recipients.

Isolation

Campaigns run on the marketing stream. Separate pools, budgets and reputation boundaries mean a campaign cannot take capacity from password resets or receipts.

Questions

Campaigns and A/B testing FAQ

+How does A/B testing work in Transmitly?

Variant B can override the subject, HTML or text independently. For a split test, recipients are assigned A or B when the audience is snapshotted and both variants send together. For an automatic winner send, A and B each receive half of a configured test sample, the rest wait, and the winning variant is sent to the remainder after the wait window.

+How is the winning variant chosen?

You choose unique open rate or unique click rate. After the window, Transmitly compares the rate among delivered test recipients, falls back to all test recipients when nothing has delivered yet, and sends the winner to the remainder. Ties go to variant A. The decision is recorded on the broadcast and emitted as broadcast.winner_selected.

+What does an open actually measure?

An open is a request for the tracking pixel. Privacy proxies and image blocking mean opens are a directional signal, not proof that a person read the message. Clicks are usually a stronger signal but are also affected by prefetching and security scanners. We never convert either into an invented inbox-placement score.

+Can I send a campaign without a winner test?

Yes. Use a normal split test to send both variants to the whole audience, or skip variants entirely. Automatic winner sends are optional and require a B variant plus the tracking setting that matches your chosen metric.

+Does marketing email share capacity with transactional mail?

No. Transactional and marketing run on separate streams with their own pools, budgets, priorities and reputation boundaries. A marketing surge cannot consume product-mail capacity, and rejected marketing traffic is never laundered onto healthy transactional IPs.

+Is there a rich editor, or do I have to write HTML?

Both. The campaign composer includes a visual editor with headings, lists, links and formatting, plus an HTML source view. Placeholders such as {{name}} resolve per contact at send time.

Want the caveats? Read A/B testing email campaigns without fooling yourself.

Send a campaign you can measure.

Import an audience, write the campaign, and run a real test send. Start free; tracking, A/B tests and winner sends are included.