Rebuilding flows after a migration: how to recreate what will not export

Contacts export. Flows do not. Every platform stores automation logic in its own shape, and the parts that matter — when a step fires, what branches it takes, what pulls somebody out — come across as nothing at all.

This is the part of a migration that takes the longest and is planned for the least, because the list is what everybody worries about and the list is the easy half.

Document before you lose access

Do this while the old tool is still paid for. Once the subscription lapses, the logic is gone and you are reconstructing it from memory and from what customers received.

For every live automation, capture:

1. The trigger, exactly. "Cart abandoned" hides a definition — after how long, requiring what.
2. Every delay, in hours, and whether it respects quiet hours or a send window.
3. Every branch condition and what each arm does.
4. Every exit condition. This is the most important one and the most commonly forgotten.
5. The audience filter applied on top — often an engagement gate nobody remembers adding.
6. Screenshots of the whole diagram, and the HTML of every email in it.

Exit conditions are the part that hurts

The copy you can rewrite. The exits are what make the sequence behave like it belongs to a shop that pays attention.

• Exit on purchase, immediately, mid-sequence. Requires the new tool to learn about the order in seconds rather than on tomorrow's sync.
• Exit on unsubscribe, obviously, and on hard bounce so a dead address does not walk the whole thing.
• Exit on entering a more specific flow — somebody who abandons a cart should leave the browse flow.
• Exit on refund or return, so a post-purchase sequence stops for somebody who sent the goods back.

The order to rebuild in

Not all at once, and not the highest-revenue one first.

1. Welcome series. Low risk, new subscribers only, and it exercises the whole path from form to send.
2. Post-purchase. Also low risk, and it proves the new tool is seeing orders correctly — which is the thing everything else depends on.
3. Win-back and replenishment. Slow-moving, forgiving of a few days' gap.
4. Browse abandonment.
5. Cart recovery, last. It has the most revenue attached and the least tolerance for a mistake, and by the time you reach it you have already validated order events three times over.

Verifying each one before switching it on

A flow that looks right in the editor is not a flow that works. Walk it.

• Put yourself through it end to end, with a real address, doing the real thing — abandon an actual cart, place an actual order.
• Verify the exit fires by buying mid-sequence and confirming the next email does not arrive.
• Check the timing against the clock, not against the configuration. Timezone handling is where rebuilt flows most often differ from the originals.
• Switch the old one off before the new one on, and confirm it is off rather than paused.

Rebuild is the only cheap moment to improve them

You are reading every flow in the account anyway, with the performance data in front of you. That combination happens roughly once every three years.

Delete the ones that never earned anything rather than porting them out of completeness. Add the holdout you always meant to add. Fix the exit conditions the old tool could not express. Migrating a bad flow faithfully is work spent preserving something you would not build today.