Changes take time to spread
A retailer changes a price at headquarters. Regional catalogues receive the update over the next few seconds, so customers in different places may briefly see different prices.
In a distributed system, a change may reach one server first while copies on other servers or in caches catch up later. Eventual consistency allows this temporary mismatch.
When delays matter
A short delay may be fine for follower counts or recommendations. It can cause serious problems when checking whether a payment succeeded or the last seat is still available. Teams decide which tasks need up-to-date data and tell users when a result is still pending.