wojciech-b thanks for the feedback 🙂 to your question - attack disruption works best when all of M365D capabilities are up and running, as it is needed to:
1. Detect the attack in high confidence
2. Incriminate the compromised asset used by the attacker
3. Contain it.
Note the above does not necessarily mean that disruption won't happen in cases of partial setup (aka suspend user can work as you pointed), but you will be less protected, as some attack scenarios will require multiple actions to effectively stop the attack.