I have 40 things to fix.
Which three actually matter?

Sort problems into four buckets.

1. Blocks the core outcome

If a meaningful number of users cannot sign up, pay, upload the thing, generate the result, book the appointment, or whatever your product exists to do, this goes first.

2. Damages trust

Incorrect billing, lost data, broken auth, confusing permissions, unexplained failures, and anything that makes a reasonable user wonder whether the product is safe belong near the top even if they are not visually dramatic.

3. Creates repeated friction

One tiny annoyance in a flow people repeat ten times can matter more than a spectacular bug on a page almost nobody sees.

4. Mostly bothers the builder

Inconsistent spacing, internal code ugliness, an animation you hate, or a feature that would be cool someday may all deserve attention later. They do not become urgent because you notice them every day.

Add the milestone filter.

Now ask what you are trying to accomplish next. A pre-launch product and a product with fifty active users can have the exact same issue and reasonably prioritize it differently.

• Before launch: prioritize the core flow, trust, recoverable failures, and basic measurement.
• After launch: prioritize what real usage shows, especially activation and repeated friction.
• Before a sales demo: prioritize the flows the buyer will actually see and the claims you need to prove.
• Before handing off development: prioritize ownership, reproducibility, documentation, and the areas most likely to surprise the next person.

Use effort only after impact.

Quick wins are useful, but “easy” should not outrank “important.” First identify the highest-impact problems. Then use effort to decide which of those you can move now.