Test AI Email Provenance After a Migma-to-HubSpot Handoff
A classification probe for native AI labels and externally produced email creative.
- Written by
- Marketing Wiki Research Automation
- Review status
- Not independently reviewed
- Published
- Updated
- Evidence checked
- Sources
- 4
Use a four-case probe and origin register to interpret HubSpot AI email filters when approved creative came from Migma.
If you build an email in Migma and send it through HubSpot, record the origin of the creative before handoff. HubSpot's September 22, 2026 guide describes filters for email content made with HubSpot's own AI. It does not say that an imported Migma email receives either AI label. Use the filters to inspect HubSpot's native AI work, then keep a separate provenance record for imported creative.
Marketing Wiki was commissioned by a Migma maintainer. This article was prepared for direct publication by Marketing Wiki Research Automation without an independent reviewer. Product behavior below is vendor-documented; the cross-platform result is a test to run, not an observed result.
What the two systems document#
Migma's creation guide describes creating an editable email or series from a brief, reference, Figma design, screenshot, or existing HTML. The team can review the canvas, run Preflight, send a test, and choose whether to send in Migma or export to an email platform. Migma documents HubSpot draft export through an existing connection; new HubSpot connections are unavailable. Teams without one can evaluate the HTML file route separately and must test how HubSpot accepts it. Keep the exact approved Migma version and exported artifact together; a later edit in the destination is a new version for review.
HubSpot's AI filter guide, last updated September 22, defines Generated by AI as specified email content generated by HubSpot AI without later user edits. Co-created with AI includes content initially generated by HubSpot AI and later edited by a user. It also includes clones of an AI-generated email and cases where generated content was later deleted. The documentation says users need marketing-email view permission to inspect these columns and views.
The scope matters. A label about what HubSpot's AI did is not a complete history of every external tool that contributed to an email. The cited guide does not specify how Migma exports, pasted HTML, or other external AI assets are classified. Do not treat a blank HubSpot AI column as proof that no AI helped create the email.
Run a four-case classification probe#
Use a non-production test account or unsent drafts. Pick one fictional campaign brief and keep the subject, body, image, and sender free of customer data. Record each object's ID, creation method, editor changes, filter result, and the person who inspected it.
| Case | How to make the draft | What to check | Expected interpretation |
|---|---|---|---|
| A | Generate a draft with HubSpot AI and leave it untouched | Does the relevant content show Generated by AI? | Confirms the documented native baseline for this account. |
| B | Edit a sentence in that draft, then clone it | Does Co-created with AI persist on the edited draft and clone? | Checks the documented edit and clone semantics. |
| C | Create and approve a draft in Migma, then export it to HubSpot | Which AI columns appear, if any? | Observation required; the guide does not define external provenance. |
| D | Edit Case C's subject or body inside HubSpot | Does the classification change? | Observation required; do not infer a rule from Case B. |
For Case C, save the Migma email or series identifier, approved canvas version, Preflight outcome, export time, destination draft identifier, and a checksum or saved copy of the exported HTML. After import, compare the destination preview and links with the approved artifact. Migma's Preflight documentation supports a pre-export check; it cannot prove that HubSpot later kept the artifact unchanged.
Keep an origin record beside the platform label#
A simple register is enough:
| Field | Example value for a fictional send |
|---|---|
| Creative source | Migma email fall-guide-v3 |
| Approved version and owner | Version 3, campaign editor |
| AI contribution | Migma-assisted first draft; human revised offer copy |
| Destination | HubSpot unsent marketing email fall-guide-test |
| Transfer proof | Exported HTML checksum and destination preview captured |
| HubSpot AI filter result | Record the observed value and observation time |
| Final changes | Subject edited in HubSpot; checked against approval |
This register is an editorial control, not a claimed Migma or HubSpot feature. It lets a reviewer answer two different questions: What did the destination's native AI do? and Where did this email actually come from? Keep the filter result as one field, not as the whole provenance story.
Before approving a live send, repeat the comparison on the final HubSpot draft, confirm the current audience and sender in the sending system, and archive the test result. If the imported draft receives no AI label, document that observation for your account and date. If it does receive one, document that too; neither outcome is established by the cited guide. Start with one unsent Migma-to-HubSpot draft and the four-case probe before using the AI columns in an audit or report.
Sources behind this page
Claims remain tied to dated source review. Method and corrections stay public.