Parallel run and cutover, one group at a time
Why moving one group at a time beats a big-bang switch, and the exact steps of a cutover with a fallback window.
A big-bang switch bets the whole support operation on one evening. A group-by-group cutover bets one team for one day, with Zendesk still there as a safety net. Here is how it works in ChaosDesk.
Plan the waves
In the Migration center, a wave is a set of Zendesk groups that move together, for example “Billing” first, then “Returns”, then everyone else. Each wave goes through the phases Planned, Parallel run, Live and, if needed, Rolled back.
Pick a first wave that is small, has a lead who likes details, and uses most of your common rules. You learn the most there.
The parallel run
During the parallel run, delta imports copy new and changed tickets from Zendesk every few minutes. The wave’s leads check, on real tickets:
- views show the same tickets in the same order;
- forms show the right fields under the same conditions;
- SLA targets and breaches match;
- macros produce the same replies.
Nobody answers customers in ChaosDesk yet. Differences are cheap to fix at this point.
Cutover day
For the groups in the wave:
- Point their support addresses at ChaosDesk.
- Run a final delta import.
- Turn off their triggers in Zendesk, so both systems never answer the same customer.
- Point the integrations they use at the Zendesk-compatible API.
- Replace the widget snippet on the pages they serve.
- The agents move over.
The fallback window
Zendesk stays read-only for the moved groups during a fallback window you choose. If something is wrong, roll the wave back from the Migration center and try again next week. Nobody has to work all weekend.
After the last wave
When every group is live and the fallback windows have passed, export what you need for your records and cancel Zendesk at your renewal date.
Teams with more than 20 agents get this planning for free. Talk to us.
Zendesk is a trademark of Zendesk, Inc. ChaosDesk is not affiliated with Zendesk.
