Many of these requirements fall into "RASP":
Reliability - Unscheduled downtime
Availability - Scheduled uptime
Scalability - Users, concurrent requests, database & network load
Performance - Response times
It can be very expensive in real dollars to optimize any one of these, and fabulously expensive to optimize all of them. Prioritize, set realistic goals, consider the dollar cost of every improvement, eg 99.9% reliable to 99.999%. Users will ask for the moon until you tell them what it costs.
Also consider how you will
test and prove any of these things. How do you prove "95% of screens complete within 2 seconds" or "99.999% reliable"?
Also consider "graceful degradation". What happens when one of many partner systems fails ... does your whole system fail or just report that one function is unavailable? What happens when one app server out of a cluster fails ... do users move seamlessly to the next one?
Any of that help?