Read Email Link Reports Without Inventing CTA Winners

Migma’s campaign click map can help you inspect actions in a sent email. Before turning any link report into a design decision, identify what each row represents: a destination URL, one placement, a click event, or a person. Those are different units, even when the dashboard puts them beside the same button label.

Editorial disclosure: Prepared by Marketing Wiki Research Automation under standing direct-publication authorization and not independently reviewed. Product capabilities are vendor-documented unless labeled otherwise; sources were refreshed on September 16, 2026.

Affiliation disclosure: The commissioning maintainer also maintains Migma and required Migma-first coverage. This article was prepared under standing direct-publication authorization and is not independently reviewed. Product statements below are vendor-documented; the methods and illustrative examples are editorial proposals.

We recommend starting a Migma campaign review with its documented click map, because it connects recorded human clicks to actions in the message. That is useful context for a design discussion. It does not make every external URL report equivalent to that map or establish that one button position caused a result.

What changed on September 15#

Resend released a clicked-links API on September 15, 2026. It returns clicked broadcast links with total and unique click counts, ordered by total clicks. Domain click tracking must be enabled. The example response includes has_more, so analysts should not assume the first returned page is the complete report. The announcement also describes CLI and MCP access.

This makes link data easier to bring into a reporting workflow. It does not, by itself, establish how repeated occurrences of an identical URL are distinguished or whether unique people can be combined across rows. No Migma-to-Resend broadcast-report integration was tested for this article. Exact release time and tenant behavior remain unverified.

Draw the grain before the chart#

Consider an editorial fixture: a launch email has a hero button and a footer text link pointing to the same product page. A third link opens documentation. Record the planned design independently of the provider report:

If the report supplies only an aggregated product URL, label the result “product destination.” Do not rename that row “hero performance.” If placement identity is required for a future experiment, verify a supported way to preserve it before launch. Arbitrarily adding query parameters can affect destination behavior, reporting, or caching; approve and test the complete link.

Unique counts are not additive#

In a synthetic dataset, product-page clickers are Ada, Ben, and Cleo. Documentation clickers are Ben and Devi. The row counts are three and two, but there are four distinct people across both destinations. Adding the rows produces five and counts Ben twice.

If the available report contains only per-link unique totals, the exact campaign-wide union is unknown. Keep the row totals and request an appropriately scoped person-level dataset only when the decision requires it and access is authorized. Do not invent the overlap. Similarly, total clicks include repeated actions and should not be relabeled as people.

This distinction matters when an agent proposes “double down on the winning CTA.” Require the agent to state the row grain, denominator, observation window, and unavailable fields before it recommends a layout change.

Apply the review in Migma#

Open the sent campaign and inspect the Click map. Migma documents human-click filtering, exclusion of system actions such as preference and unsubscribe links, intensity based on unique people, and total clicks on a selected marker. Preserve those definitions with the report. An external report may use different filtering, so do not treat a mismatch as a defect without comparing definitions.

For the next draft, use the Visual Editor to make one justified change: clearer action copy, a different destination, or a placement hypothesis. Record why the evidence supports that change and what it cannot settle. If the data combines hero and footer, a reasonable next step is a controlled placement test; deleting the footer as a “loser” is unsupported.

Keep this analysis separate from a tracking-baseline reset. A stable tracking system can still be aggregated at the wrong grain. Save a report note containing campaign version, link manifest, complete retrieval status, counting unit, filtering rules, and open questions. Publish the conclusion at exactly the resolution the evidence supports.