Feedback from everyone is one very important part, but we also look at hard data and how a change would affect every other part of the game.
Honestly Gina, I've enjoyed many conversations with you and various Devs since Beta. But the claim that you're using "hard data" to determine the affects of various changes is laughable considering the current state of your game.
*****
What hard data did you examine when balancing the CP system? Do passives like unchained look balanced to you?
What hard data did you examine when implementing Battle Spirit, especially in relation to class defining skills like Dragon Blood?
What hard data did you use when implementing a no CP campaign? Did you adjust Streak / Dodge penalties to compensate for 0 CP PvP?
What hard data did you use when implementing crutch sets like Shield breaker? Did you adjust this set after nerfing sorc shields?
What hard data did you use when setting shield duration at 6 seconds? Are your matrices showing an acceptable number of MagSorcs in Cyrodiil?
What hard data did you use when implementing proc sets like Valendrith and Viper? Do you feel these sets increase or decrease the importance of player skill in ESO?
What hard data did you use when buffing heavy armor? If your data is reliable and / or your ability to interpret this data is satisfactory, why are HA sets like Black Rose so popular among damage dealers?
What hard data did you use when buffing Soul Assault? Is another long-range, undodgeable beam encouraging or discouraging smart counter-play in ESO's PvP?
What hard data did you use that allowed gap closers to exist in their current state? Has banning players for using a mechanic you developed helped to address the underlying mechanical issues?
What hard data did you use to balance Masters and Maelstrom weapons for Magicka and Stamina users? Do you feel that Magicka users benefit from their vMA weapons as much as Stamina users do from theirs?
What hard data did you use when deciding that multiple poisons should apply on one person? Do you feel zerglings need additional advantages to be successful in 20 v 2 situations?
What hard data did you use when implementing Rally and Vigour? Do you feel stam builds need a burst heal from their offensive weapon (Rally) and a passive heal (Vigour) whose ticks surpasses a DKs burst heal?
What hard data did you use when deciding that Bound Armor would remain a toggle? That Storm Atronach would be the only Atronach in the game that can be CC'd?
What hard data did you reference that led you to decide that Streak should be the only CLASS DEFINING SKILL in the game that punishes the user for casting it more than once? Why is the same not true for other class defining skills like cloak, BoL and Wings?
*****
I hope you understand why the community might have a tough time understanding your use of "hard data." The current state of ESO leads me to believe that either:
a) you don't use hard data, or
b) you do use hard data, but lack the competence to use it effectively, or
c) you do use hard data and have the competence to use it effectively, but lack the monetary incentive to do so
"Men are more important than tools. If you don't believe so, put a good tool into the hands of a poor workman."
- John J. Bernet