We are 20 +/- 1 degC, but that's only because we are primarily a mechanical lab, which is strongly linked to 20. It's more than a little annoying that there is a disconnect between mechanical and electrical world because all my electrical references are externally calibrated at nominally 22 - 23, so I need to add uncertainty to allow for my lower operating temp ☹️
I will eventually overcome it by dividing the lab and running different conditions, but it is a shame to need to do so, as it seems somewhat arbitrary.
As Berni said, I think good control is most important. In an ideal world we would match the conditions to those of the end user, but we don't often know that (especially for field equipment) and it would be impractical to change lab temp often. As long as we inform the customer of the test conditions during contract negotiation and on the report, they can consider the implications.