The 8-factor scoring rubric

Before you write a draft, you evaluate. Evaluation takes six seconds, runs on a fast model, and gives you a composite 0–100 score plus eight axis scores. The point isn't to find the "best" idea. The point is to kill weak ones early.

The 8 factors

1. Hook strength — does the opening line stop a scroll?
2. Specificity — are there concrete numbers, names, details?
3. Audience fit — does your target reader actually care?
4. Distinctiveness — is this something 100 other people haven't already posted?
5. Timing — relevant now, or stale?
6. Emotional charge — does it provoke a reaction?
7. Payoff potential — does the body deliver on the hook?
8. Shareability — would readers forward it?

The composite score

Factors weight into a 0–100 composite. Default weights are roughly even, with hook strengthand specificity weighted slightly higher (they correlate best with virality in our telemetry). You can adjust weights per app — see customizing factor weights.

The verdict

Each evaluation returns a verdict: kill, refine, or ship.

• kill — composite under 40. Don't write it. File it or delete it.
• refine — composite 40–70. Rework the hook or tighten the specifics.
• ship — composite over 70. Proceed to drafting.

Why it's not just "pick the highest score"

See why we killed score-above-all. Briefly: some posts are meant to start conversations (lower shareability, higher emotional charge) while others are data drops (higher payoff, lower hook). Scoring with a single number loses this context. The axis scores are more useful than the composite.

Reading the output

Evaluation produces a structured object we render as a small bar chart plus a one-sentence reason. Hover any factor to see why the model scored it that way. Save the evaluation — it becomes part of the idea's history.