UX isn't just about doing what the user wants, or even what the user needs. Good UX requires taking many elements into consideration.
Business Requirements - Does this requirement help achieve any measurable business benefit? Does giving every customer a free cup of coffee directly increase the number of actual sales you make?
Technical Limitations - What can technically be achieved within the infrastructure / environment / platform / framework. If you have an iPhone app, then it doesn't really matter if all the users want 3D holograms. It ain't going to happen.
Budgetary Restrictions - How much money does the client have available? Do they have available resource (i.e. designers, developers, project managers...) to even produce it? - 1 hour delivery by helicopter is a great idea. I doubt many (any) companies can afford that though.
User Benefit - Does the requirement give any measurable benefit to the end user? - Changing the colour from red to burgandy might be something many people suggest. But who is that actually helping?
A lot of this is around measurability. Sure, it's nice to have some cool feature built on your website, but if there's no way to gauge the impact of that feature then you might have wasted your time building it. Did it bring more people to your site / app? Have you gained more sales?
If the requirement has a measurable benefit to the business, if it is cost-effective to produce, can be developed within the existing environment and gives a benefit to the end user then that makes it a valid requirement. If it's just something cool that lots of users want, then it's less so.
download bmml source – Wireframes created with Balsamiq Mockups
The closer to the centre of this Venn (Euler?) diagram you can get, the more important that requirement is to get live.