My second (full-time) job, I worked as a Document Controller. In large projects you need people to track documentation, build a library and disseminate them to the interested parties.
Not long after I started, one of my colleagues introduced me to the concept of 'benevolent inaction.' Which is to say, leaving a task for a while, because it often saves you time in the long run.
Our main duty was to take documents dropped off by engineers, copy them, then distribute and/or file those copies. By copy I mean photocopy. Depending on the size of the document and the number of teams it impacted, this could be a lot of work. We had an in-tray for processing and engineers would simply leave the documents with us.
As an eager newcomer, my natural instinct was to process the waiting pile as fast as I could. I didn't want to be responsible for holding up a project or causing expensive alterations because someone didn't get a document in time. The thing is, engineers aren't perfect. Not often, but regularly, an engineer would sheepishly reappear after leaving his document and take it back to make amendments. They'd forget something, overlook something, maybe they'd spot a typo.
If that happened after the document had been processed, it meant a new revision and another round of copying. So I learned to let the documents wait a while, not too long, but long enough to eliminate the sort of things we seem to remember only after we finish.
It's something I have learned to do in all my jobs. Obviously there are some things that blatantly can't wait, those that are impacting people right now, but it's also very easy to come up with an idea and ask someone to do it without thinking about the work involved, let alone whether you actually need it.
It could be an idea that quickly gets replaced, or a fad that is hot this week, but not next, or the result of an emotion that dies down with time. You have to make a call on whether the task you've been given can be left, which is sometimes a challenge. There's no hard and fast rule to identify them. As long as it's not business critical or impacting people right this second, you're probably safe. The old faithful 'when do you need it by?' can be a helpful way to gauge the priority.
If someone really wants the job done, they'll chase you on it, or at least follow up to check progress. If not, you just saved yourself some time that can (probably) be better spent elsewhere.