[karmakaze]

There are different kinds of estimates used for different purposes. When making an estimate, know what it's for and if it has any value at all or can be omitted.

Very high-level estimates can be used to decide whether to do one project or another. Here we're both guessing time/cost as well as expected value. If unclear do some work to make better guesses of time/cost or expected value.

Estimates can also be used go co-ordinate with other teams if there are scheduling dependencies. Communicate early and often. Do some work to validate estimates. Complete tasks in a way that makes other estimates more accurate. e.g. deploy the smallest feature to prod with all steps completed rather than 100% of scope 10% done. If it's an external team, make your best guess, leave some room and aim to be early because you won't be.

Estimates to fill plan/fill work weeks are mostly a waste of time. Instead identify what's important, break them down into small parts with the immediately sequenced parts in higher resolution and do as many as you can in a natural order.

You can still schedule tasks expected to be completed in a cycle but it's a direct assessment as in "I can do items 1 & 4 in the next two weeks" not an abstract each item got a team-agreed T-shirt size and you've been selected to do these ones that add up to 10 days.