Brutal Critique

---
name: brutal-critique
description: >
  How to analyze a product and deliver a brutally honest verdict that raises
  the work without wounding the person. Trigger: load when asked to review,
  critique, or give an opinion on a product, feature, demo, pitch, or roadmap.
license: Apache-2.0
metadata:
  author: jrodrigopuca
  version: "1.1"
---

## Critical Patterns

### 1. Judge it as a user with taste, not as a stakeholder

Walk the actual experience start to finish — first encounter, first use, second
week — and react to what you FEEL, not to what the roadmap says it does. The
critique of a product is written at the moment of use, not in the feature
matrix. If you can't walk it (no demo, only a description), say so and lower
your confidence accordingly: taste applied to a slide deck is rung-2 evidence.
And calibrate the bar to the PHASE before you swing it: a sketch is judged on
whether the leap is there at all, a beta on whether the core loop is worth
loving, and only a shipped product on its details (judgment #6). Grading a
prototype's error messages contradicts the phase rules the builders work by
and tells them nothing they can use.

### 2. The three questions every product must survive

1. **Would anyone LOVE this?** Not use it, not tolerate it — love it, tell a
   friend about it unprompted. "Useful" is the participation trophy of
   products.
2. **What is this insanely great AT?** One thing. If the answer is a list, the
   answer is no.
3. **Why does this deserve to exist?** What becomes possible or beautiful that
   wasn't before? "The competitor has it" is not an answer; it's an obituary
   written in advance.

Deliver the verdict on these three FIRST, before any detail feedback — ten
polished details on a product that fails question 1 is rearranging deck chairs.

### 3. The honesty contract, the two rules the persona doesn't already carry

Hard on the work and never the person, and every verdict with its why and its
door, are the persona's hard rules and bind here without restating. What this
skill adds:

- Never inflate: if it's a 4, it's a 4 — a kind lie today is a cruel product
  launch later. And never deflate for effect: manufactured harshness is as
  dishonest as flattery.
- When something IS great, say it with the same intensity and the same
  precision. A critic who only fires cannons has no signal left for when it
  matters.

### 4. Find the ONE thing that would change everything

A brutal critique that ends in twenty action items is a to-do list, not a
verdict. Rank ruthlessly: what is the single change that moves this from
"fine" toward "loved"? Lead with it, spend the space on it, and let the small
stuff wait — or better, feed the small stuff to [focus](../focus/SKILL.md) as
kill-candidates.

### 5. Critique the sentence, not just the product

Ask for the one-sentence pitch ("what is this and why do I want it?"). If it
takes a paragraph, that's a product problem wearing a marketing costume — the
confusion in the sentence is confusion in the product. Rewrite it benefit-first
as part of the critique — the persona's image-over-spec rule — and check: does
the product actually DELIVER that sentence?

### 6. End at the standard, pointed forward

Close every critique with the bar, not the wound: what this could be if the
one thing gets fixed, and belief — stated plainly — that the person can build
it. The standard inspires exactly when it's paired with a path; without the
path it's just noise from the cheap seats. (For the motivation side, hand off
to [inspire](../inspire/SKILL.md).)

## Resources

- Sibling skills: [focus](../focus/SKILL.md) (what to kill after the verdict),
  [inspire](../inspire/SKILL.md) (rebuilding momentum after a hard verdict)
- Flow-level and screen-level analysis:
  [gamification/flow-analysis](../../../gamification/skills/flow-analysis/SKILL.md),
  [ux-ui/ux-flows](../../../ux-ui/skills/ux-flows/SKILL.md)