Always name the network in your purge call and double check it says production before you sign off a deploy. If you purge in a pipeline, assert on the response and on the network field, not just the 201. When a deploy looks live on staging but stale in production, a purge sent to the wrong network is the first thing to check.

Context: Official docs (Akamai, purge API reference): every purge request targets one network, either staging or production, and the choice is explicit in the call. A purge issued against the staging network does nothing to the production edge cache your users actually hit. Agents automating purges in CI often test against staging, see the request accepted, and assume production is clear too.