Answer
A useful exercise is to price out downtime by department rather than treating the whole business as one number. An hour of downtime during a shipping cutoff window might cost missed carrier pickups and expedited freight charges the next day, while the same hour in finance during a non-critical period might cost almost nothing. Walking through this by function, and by time of day and month, since month-end close or peak season changes the calculus, gives a far more defensible downtime tolerance than a single company-wide RTO figure pulled out of a planning meeting.
Once that tolerance exists, it should drive concrete technical decisions rather than sit in a binder. A five-minute tolerance argues for synchronous or near-synchronous journal-based replication with an automated or semi-automated role swap; a same-day tolerance might be met with a well-tested backup and restore process instead of the expense of full HA. Buyers should ask vendors to map their product's actual measured failover time, not a theoretical best case, against the business's real tolerance, and should ask what staffing is required to execute a role swap at 3 AM on a weekend versus during business hours, since many HA failures are really staffing failures rather than technology failures.