You're ignoring the counting stats just as much as he's ignoring the analytics. Which is more important?
The use of counting stats as a predictive tool is obviously flawed, but everyone accepts that. There is variance, and totals can change from year to year without any change on the part of the player. But on the other hand, the use of analytics in discussions like this is far from "proof." Using a statistic as support for a claim involves an implication that the statistic has a certain meaning with respect to actual, real hockey. Whether that meaning is valid is very much up for debate, and analytics as a field has a long way to go and some serious issues to fix.
Chief among those issues is that generating shots is not the be all and end all in hockey. In addition, scoring is a talent, and defending in your own zone is a talent. Watching Bennett, he didn't have nearly enough shifts last year where his line outplayed the other line. He didn't generate nearly enough high quality scoring chances, and he got beat in his own end leading to him taking penalties. He was not, most nights, a positive difference maker for the team. That needs to change somehow. Pointing to those numbers and suggesting essentially that if he plays identically next year to how he played last year, he'll get more goals by law of averages does not jive at all with what we watched game to game last season. He has a lot of improvements to make.