01 The symptom
Pages take thirty seconds. Reports time out. Lightning components fail to load. Attachments do not upload. The mainland team quietly stops using the CRM and starts keeping things in spreadsheets, and nobody tells the CRO until the pipeline data stops making sense.
None of this is a Salesforce fault and none of it is fixed inside Salesforce. It is cross-border latency, packet loss and a large number of round trips against an origin on the wrong side of the gateway — the same problem as any foreign-hosted application, made worse because the CRM is unusually chatty.
We have been working with Salesforce since 2014 and on the China access problem since 2017, which is a fairly narrow intersection of two things.
The same pattern applies to every SaaS tool your mainland team uses.
Salesforce is the one people notice first because it is where revenue lives. The ticketing system, the wiki, the code host, the design tool and — now — whatever AI assistant your company standardised on all have the same problem and the same class of solution.