There's a huge problem with using "x Max" public data win rates without nuance because they only count games where players have reached enough levels to actually reach rank 5 of the ability. I spot-checked this once for Seraphine and it was somewhere between 1% and 2% win rate inflation at Support, less so for Mid/Bot.
If you want a better public data resource for that, just look at what's ranked up at champion level 5. You're much less to have the above bias and it'll be closer to the truth (though it has its own biases). Using lolalytics as an example, about as many players go 3W -> 5E (level 10 required) as go 5E by 9. This means you're heavily inflating "E-max" win rate builds because being 1 level higher biases toward being on the winning team.
E max Janna is like 50.5 and W max Janna is closer 53. I believe Janna is a completely reasonable champion in the 51.5-52 range if she's not severely MMR skewed.