Turn Vague Feedback into Tasks
Make “make it better” testable
Translate the vague feedback below into executable work. First list three specific problems it could mean, each with an observable signal. Then ask me the three questions I have to answer — goal, current state, and acceptance criteria. Then propose a default plan: what changes, what does not, how you will verify it, and the level of effort. Do not start rewriting, and do not treat “optimize” as “redo”. [Vague feedback] [Current version/data]
How to use
Ask “what would better look like” before “what is wrong now”. If neither has an answer, the feedback is a feeling, not a requirement.
Good for
- In review, your manager says “make this page better” without saying what is wrong, and the designer is already on the third revision.
- A customer says the report should “feel smoother”, but cannot say whether they mean speed, definitions, or the number of clicks.
Not for
- It is already known to be a performance problem with monitoring data — go find the bottleneck; no need to clarify requirements.
- A senior person gave one directional sentence — align with a cheap concrete option before asking ten questions.
Example
Input
Boss in the group chat: Optimize this campaign page — conversion feels low. Current conversion is 2.1%, the industry average is around 3%. The page runs, top to bottom: banner, rules, product list, sign-up button.
What you get
List three possible problems: the offer is unclear (signal: first-screen dwell under three seconds), the rules are too long (signal: few people scroll to the product list), or the button is too far down (signal: clicks cluster at the page bottom). Ask for the target conversion rate, what may change, and the launch date. Default plan: fix the first-screen offer and button placement, collapse the rules, leave the product list alone, run an A/B for a week and decide once the sample is sufficient — roughly one day of design and two days of engineering.
Common mistakes
- Replying “sure, I'll optimize it” — they think you understood, the next version still feels wrong, and a round is wasted.
- Reading “optimize” as “redo everything” — you overturn the structure without locating the problem, lose the comparison, and throw away what already worked.
Try tweaking it like this
- Add “give me three directions, each with a one-line expected effect”, for someone who cannot articulate the problem but can recognize a fix.
- Add “start with the smallest experiment that changes one thing”, so data settles the argument on conversion.