After a migration: what to check on day one

A migration that goes wrong rarely fails loudly. Mail still sends, the dashboard still populates, and the problem is a suppression list that did not import or an exit condition that does not fire. Both are silent for days and expensive by the time somebody notices.

This is the list to work through on the first morning, in this order, because the order is roughly the cost of getting it wrong.

First: prove the suppressions arrived

The highest-consequence failure in any migration is mailing somebody who unsubscribed.

1. Count the suppression list in the new tool and compare with the export from the old one. They should match, not approximately.
2. Spot-check ten known unsubscribers by address. Counts can match while the wrong rows imported.
3. Send a test campaign to a segment that ought to exclude them and confirm the count drops by the expected amount.
4. Check hard bounces came across too, not just unsubscribes. These are frequently a separate export and frequently forgotten.

Second: confirm mail is actually being delivered

• Read raw headers of a real message at Gmail, Outlook and one more provider. SPF, DKIM and DMARC all passing, with DKIM aligned to the From domain.
• Check the inbox, not the sent log. Then check the spam folder, which is where a migration problem shows first.
• Compare click rate to the old tool on a comparable audience. A material drop with the same list means placement, not content.
• Confirm the unsubscribe link works and points at the new tool. A footer link still pointing at the old platform will stop working the day you cancel.

Third: confirm orders are arriving, fast

Everything downstream depends on this and it is easy to verify.

1. Place a real order in the shop and watch how long it takes to appear. Seconds, not minutes.
2. Check the value, currency and tax treatment match the shop exactly — not approximately, and not the order total where you meant goods total.
3. Refund it and confirm the refund arrives too. Refunds are handled by a different path and are commonly missed.
4. Confirm the order exits a flow. Enter cart recovery, buy mid-sequence, and verify the next email does not arrive. This is the single most valuable test on this page.

Fourth: walk each flow for real

Not read the configuration — walk it, with a real address, doing the real thing.

• Trigger each flow genuinely: subscribe, abandon a cart, place an order.
• Check the timing against a clock, not against the settings. Time zone handling is the most common difference between a rebuilt flow and the original.
• Verify every exit condition individually.
• Confirm the old tool's equivalent is off, not paused, and that nobody received two of anything.

Then watch two numbers for a fortnight

Most of what remains shows up as a trend rather than an event.

1. Complaint rate per send, against the 0.1 % line. A rise after migration usually means suppression or audience problems rather than content.
2. Revenue per recipient, compared with the same campaign type before the move. This is the only cross-tool comparison that means anything.

And before cancelling the old platform: export contacts, suppressions, campaign history, every template's HTML, screenshots of every automation, your form wording, and remove the old provider's SPF include and DKIM records. Then keep those files somewhere you will find them in two years, which is when somebody asks about a consent given before the move.