This is general suggestion. It would be more helpful if you share VT results of samples that were faster than Behavior Shield.
That way they can study the samples