# Estimator of a normal mean: practical recommendation?

I have a sample of $n=370$ $5$-dimensional observations which can be reasonably well modeled using a multivariate normal distribution. I want to estimate the mean using this sample, which will be part of a report, and I came across the Stein's paradox. Given that the James-Stein estimator is biased and the sample mean is unbiased, I am unsure on whether I should report the Sample Mean or the James-Stein estimator of the mean (or both?).

What is the practical recommendation in this case?