DNS propagation is the delay between changing a DNS record at your authoritative nameserver and every resolver on the internet serving that new value. It is not a single switch; it is thousands of independent caches expiring at their own pace.
How resolvers cache
When a resolver looks up your record, it stores the answer for the length of the record’s TTL (time to live), measured in seconds. Until that timer runs out, the resolver answers from its cache without asking your nameserver again. A record with a TTL of 3600 can therefore be served from cache for up to an hour after you change it.
Why locations disagree
Different resolvers looked up your record at different moments, so their caches expire at different times. During propagation you will see a mix of old and new answers depending on which resolver each location uses. This is expected and clears on its own.
Planning a change
Lower the TTL well before a planned migration — for example to 300 seconds — so that when you make the switch, caches everywhere expire within five minutes. Once the change has settled, raise the TTL again to reduce load and cost.
Nameserver changes are different
Changing your nameservers (the NS delegation at the registrar) depends on the parent zone’s TTL, which is often 24–48 hours and outside your control. Plan those changes with more headroom.