This is very much a "How high is up?" question.
I have an AFF A700, and a FAS 9000 with only ESX hosts attached running Ontap 9.4P1. The A700 has all efficiency functions enabled, the FAS9000 is only using inline compression/compaction. Both arrays are delivering block level services via FC, with no filer protocols in use.
Some guests have data that has already been significantly compressed/deduped, and for these even 1.15:1 is reasonable.
On some other guests I have seen 2.5:1 just with compression/compaction.
Those numbers are without snapshots or cloning.
I look at efficiency with regard to placing things on the flash or spinning disk unit. If it doesn't need high IOPS/low latency, and it isn't very efficient, it goes on spinning disks.