In my experience, the VO2Max estimator for Garmin tends to produce higher estimates when doing quite slow runs and gradually building up speed, whereas intervals sessions will lead to a lower estimate. From reading their white paper they fit a curve to the heart rate reserve vs pace relationship, but I think they account for heart rate lag using population averages, which means they'll often think (for me at least) that I need a very high heart rate for a slow pace after intervals which biases the estimate down. If one starts slow then builds up speed gradually, the aerobic system is more warmed up which leads to less elevated heart rate. Also, keeping the run short-ish will lead to higher estimates since heart rate drift seems not to be taken into account.

All considered, the variance is quite high which in my experience does not make it that useful to track relative fitness changes.