What metrics are you having trouble with?
When it comes to memory, usage and workload are very different. Usage is (at a high level) the amount of resources the VM's on a host or in a cluster are staking a claim on. Workload, which is more like Demand+Hypervisor Overhead, is the actual amount of reads and writes to memory the VM is making to the underlying physical host.
Whether or not you are breaching thresholds or need to add or remove resources is really based upon what you tell vROps to tell you. Are your policies set to use Allocation or Demand or both? If so, at what percentage of workload, with how much buffer, and with what amount of overcommit?
Are you using the default policy? If so, this is problem number one with your entire vROps config.