Loops Email Tests: Freeze the Creative Before Launch

An email experiment becomes hard to interpret when the creative changes halfway through. A button gets rewritten, an image is replaced, or a product page starts showing a different offer. The experiment still has two branches, but the team no longer knows which version each reader actually saw.

Prepare the alternatives in Migma, agree on the exact difference, and freeze the approved creative before connecting it to a Loops experiment. Keep a small version record so a later edit does not quietly change the question you are testing.

This Migma-published guide describes a practical creative workflow. It is not an independent benchmark or a promise of better results. The Loops and Migma documentation cited below was checked on October 16, 2026.

Write the test question as a sentence

Start with a decision your team could make afterward. “Does explaining the next setup step above the button help new administrators complete setup?” is more useful than “Which email wins?” The first question tells the designer what may change and tells the analyst what outcome matters.

Decide whether you want to isolate one difference or compare complete creative directions. Both can be useful. If the headline, image, and button all differ, describe the result as a comparison of complete versions. Do not later attribute it to button wording alone.

Loops' A/B testing guide covers test variables, planned outcome windows, and uncertainty. This article focuses on the creative handoff that must remain stable while the experiment runs.

Build the Migma pair around fixed facts

Create the baseline and alternative from the same reviewed product facts and brand guidance. Use Migma to prepare and visually inspect both versions. Keep the approved sender, destination, offer, audience assumptions, and required footer content in the brief.

For an illustrative setup email, version A might put the setup explanation below the button while version B puts that same explanation above it. Keep the wording, destination, and image unchanged. This is a proposed test design, not an observed result.

Use this small worksheet before approval:

Keep the record somewhere your team already works. A document with links to the approved previews is enough. The point is to make a change visible, not to create a complicated approval system.

Check the imported versions before enrollment

Run Migma Email Preflight on each version. Review links, writing, inbox previews, and layout. Fix material problems before treating the pair as ready. Checks reduce avoidable errors; they do not establish statistical validity or guarantee inbox placement.

Download reviewed HTML using Migma's export workflow, then have the sending owner implement the versions in Loops using the supported approach for that account. Do not assume an automatic Migma-to-Loops connection. Open and test the resulting messages in the destination.

Compare them side by side against the approved Migma previews. Look beyond the obvious headline. Check preheader text, fallback values, tracked destinations, image dimensions, and required footer behavior. A shared accidental change can make both versions wrong even when they still look different from each other.

Understand what the Loops control branch means

The current Loops experiment documentation describes experiments inside Workflows, with variant branches and an optional control branch. Each branch must contain at least one email before the workflow starts.

That control is an email baseline. It is not automatically a group that receives no email. The documentation also says that, without a control branch, contacts outside the selected experiment sample exit at that node without receiving its emails. Have the workflow owner verify the actual setup before interpreting a creative comparison as a no-send study.

Keep creative review and experiment configuration distinct. Migma helps you prepare the reviewed alternatives. The Loops workflow owner checks enrollment, branch settings, filters, timing, and sending behavior. Your analytics owner checks the agreed outcome and observation period.

Treat a mid-test correction as a recorded change

If an urgent factual error appears, protect the reader first. Tell the experiment owner exactly which version and destination are affected. Record when the correction happened and whether any contacts had already received the earlier message.

Loops documents pausing a workflow before editing experiments. It also notes that paused sending queues contacts and that scheduled messages are sent when the workflow resumes. Review that behavior with the owner before resuming; do not assume a pause erases queued work.

A cosmetic preference can usually wait. A wrong deadline or broken action needs a deliberate decision about correction and analysis. Do not blend materially different creative into one result without recording the limitation.

After the planned analysis, save the selected version and the reasoning for the next brief. Migma's own campaign results documentation likewise describes a reviewed decision, not an automatically chosen A/B winner. Bring a specific lesson into your next Migma draft, while keeping the evidence and its limits attached.

Frequently Asked Questions