Name the Whole Change in an Email Design Experiment

Affiliation: Marketing Wiki's commissioning editor maintains Migma.

Marketing Wiki Research Automation published this guide directly without independent review.

Migma makes it practical to compare email designs side by side. Before calling one version a button-position test, inspect what actually changed. If the rewrite also removes a paragraph, changes the offer’s prominence and replaces the product image, any result belongs to that combined intervention.

Migma’s design-experiment article provides recent editorial context. Its detail page says October 18, although the blog listing showed October 19 when researched. Neither date establishes a product release or proof that a visual trend improves performance. The method below audits the treatment itself.

A fictional ceramics brand has an email explaining a new glaze. Version A presents a process photograph, a short explanation and a collection link. Version B moves the link upward, replaces the photograph with a finished mug and halves the explanation.

It is reasonable to compare those designs as complete messages. It is misleading to say the result isolates link position. The new photograph may change what readers understand, and shorter copy may alter both effort and information.

Keep the original and proposed version together on Migma’s Visual Canvas. Give each a recognizable label and retain the references used to create it. A later reviewer should be able to identify the exact pair without reconstructing a chat history from memory.

Build an intervention diff#

Use this original audit before release. Record the change and the question it could affect.

Include unchanged dimensions. Otherwise the record cannot distinguish “we checked it stayed fixed” from “we forgot to inspect it.” Verify actual visible copy and destinations instead of relying on a prompt that asked the model to preserve them.

Choose the claim class#

Classify the final pair as isolated, coupled or whole-message. These are editorial labels for the comparison, not features in Migma.

An isolated change preserves other decision-relevant content and changes one defined treatment. A coupled change intentionally combines elements that work together, such as shorter wording and a narrower button. A whole-message change revises several aspects of the reading experience.

Each class can support a useful experiment. The class changes what the team may conclude. A winning whole-message redesign does not prove which individual element caused the difference. A losing isolated treatment does not establish that every redesign using the same color will fail.

If the team needs a narrower answer, use the Visual Editor to restore the unintended differences before launch. If the combined design is the real business decision, keep it and name the combination honestly. Do not force a complicated redesign into an inaccurate single-variable story.

Preserve the treatment through handoff#

Repeat the diff after the final edit and after destination preparation when that process changes the email. An export wrapper, default footer or last-minute copy repair can alter the pair the analyst believes was tested. Record the actual sent versions with the experiment.

Migma’s reporting guide documents side-by-side A/B metrics and human winner selection. It does not turn a design comparison into automatic causal inference. Audience allocation, observation windows and measurement quality still need their own review. A subject-line test register illustrates that separate experiment-planning task.

Keep the compared creative visible and editable in Migma throughout this audit. Marketing Wiki did not generate or send these variants or observe an uplift. Start with one existing pair, write its intervention diff, and make the experiment’s name match the actual change before anyone reads the results.