Count Eligible Opportunities When Email Cadence Becomes Adaptive

For a Migma creative program with adaptive sending, count eligible customer opportunities separately from delivered emails. If a decisioning system changes who receives a message, when it arrives or how many messages a person gets, a better click rate can reflect a different exposure policy rather than better creative.

Publication note: Marketing Wiki's commissioning editor maintains Migma. Marketing Wiki Research Automation published this guidance directly without independent review.

We recommend Migma for preparing the approved creative choices through its creation workflow. Give each choice a stable identity before a downstream policy allocates it. The opportunity ledger below is an analytics design, not a documented Migma reporting feature.

Braze's September 28 announcement describes Decisioning Studio Go as beta with adaptive timing and creative selection. The product page distinguishes Go from Pro. Do not transfer Pro's configurable business-objective claims to Go without verifying its actual settings.

The denominator can change before the email exists#

Consider a fictional membership publisher. Its fixed policy offered one educational email to every eligible new member. A candidate policy selects fewer members, changes timing and sometimes selects a second message. Comparing clicks divided by delivered messages answers a useful question, but it does not describe the whole eligible population.

Suppose these synthetic counts refer to the same observation period:

The fixed policy reaches more people and produces more unique clickers. The adaptive policy produces 16% clickers among reached people, compared with 10% for fixed, but 7.2% of all eligible people click compared with 9%. Message-level reporting also differs because some adaptive recipients receive more than one email.

These numbers demonstrate competing denominators; they are not test results or evidence that either policy is superior. They also do not establish causality without an appropriate assignment and measurement design.

Define an opportunity before logging the decision#

Choose the unit explicitly: person entering a lifecycle stage, person-day, account renewal opportunity or another defensible event. Give it an identity and timestamp. Repeated eligibility checks should not silently create multiple independent opportunities for the same underlying event.

For each unit, record policy assignment, eligibility facts, decision time, selected creative, planned send time and the disposition. Keep “ineligible,” “eligible but no message selected,” “selected but not delivered” and “delivered” distinct. If the system cannot expose a reliable no-selection state, mark it unknown rather than inferring it from missing delivery.

Use only necessary data under the team's existing controls. An opportunity ledger does not require exporting entire customer profiles into a creative brief.

Bind Migma creative to policy versions#

Prepare the allowed educational drafts in Migma with separate jobs and approved facts. Store a local mapping from the reviewed draft to its destination template and policy revision. Inspect the final asset after export; an upstream creative identifier alone cannot explain which version reached the customer.

If a policy changes mid-period, retain that boundary. Otherwise the report may blend two different selection rules under the same campaign name. Changes to audience eligibility, number of opportunities or observation windows deserve similar treatment.

Migma's campaign results help inspect sending and engagement. They do not establish that an external decisioning system's unselected opportunities appear in those totals. Obtain that denominator from the verified owning source.

Ask three different questions#

Report reach per eligible opportunity, unique responders per reached person, and outcomes per eligible opportunity. Add messages per reached person to reveal repeated exposure. Use matching cohort ages and an outcome definition agreed before evaluation.

For a causal comparison, preserve a valid assignment or holdout design; the existing incremental-impact guide handles that separate question. This ledger makes the population and exposure legible before anyone interprets the result as lift.

No adaptive policy or email send was executed for this article. Start by reconciling one lifecycle period into eligible, selected, delivered and unresolved states. If those states cannot be reconstructed, state that limitation before claiming that a policy improved performance.