offer-testing

I spent some time with your skill and noticed the grading came in at 64/100 — there's solid foundation here, but the structure feels like it's fighting against itself a bit. The core idea of offer testing is useful, but I'm curious about the choices you made around when and how to surface those options to users.

Links:

• GitHub
• Your agentic skill in the SkillzWave marketplace

The TL;DR

You're at 64/100, solid D territory. This is based on Anthropic's progressive disclosure architecture principles and skill utility standards. Your strongest area is Spec Compliance (12/15) — the YAML frontmatter is clean and technically sound. But Utility is dragging you down hard at 9/20, which makes sense because the skill reads like an outline instead of a guide users can actually execute.

What's Working Well

• Valid frontmatter structure — Your metadata follows conventions correctly with proper hyphen-case naming
• Clear 5-step framework — The progression from hypothesis through analysis shows you've thought about the testing flow
• Objective tone — No marketing fluff, just instructional content that gets to the point
• Grep-friendly formatting — Clean enough that reference tools could parse it without breaking a sweat

The Big One: Missing Actual Templates and Examples

Here's what's killing your utility score: you mention templates and examples but don't provide them. You say "Experiment brief (variable, control, variant, KPI, sample size, duration)" but there's no actual markdown template users can copy-paste. Same with results reporting.

Why this matters: A marketer reading your skill has the framework, sure, but when they sit down to actually run a test, they're googling for a template anyway. You've created demand but not fulfilled it.

Concrete fix: Create references/templates.md with 2-3 actual templates (experiment brief, results report, prioritization matrix) users can steal directly. Add references/examples.md with a worked example — like an A/B test on email subject lines showing control/variant setup, sample size calculation, and results interpretation. That's probably +6-7 points right there.

Other Things Worth Fixing

1. Weak search triggers — Your description says "designing copy experiments" but nobody searches that. Add specific terms: "A/B testing", "split test", "statistical significance", "multivariate testing", "email subject lines", "CTA optimization". That's +3 points in discoverability.

2. Statistical guidance is hand-wavy — You mention "chi-square, z-test" but provide no thresholds, sample size calculators, or confidence intervals. Add a references/statistical-analysis.md with formulas and tool recommendations. +4 points.

3. No validation checklist — Where's the "before you launch" checklist? Before deploying a test, users should verify: tracking pixels work, control/variant render correctly, audience split logic is sound. Add a run-check-fix pattern. +3 points.

4. Inconsistent voice — Mix of noun phrases ("Hypothesis – statement...") and imperatives ("Limit to one variable"). Pick one and stick with it. Minor but adds polish.

Quick Wins

• Add references/templates.md with copy-paste templates (+6-7 pts)
• Expand description with specific search terms: A/B test, split test, statistical significance, multivariate (+3 pts)
• Create references/examples.md with a worked scenario (+5 pts)
• Add validation checklist before launch (+3 pts)

Hitting those four moves gets you to 80+, easy. You've got the foundation — just need the actual scaffolding users can climb.

Checkout your skill here: SkillzWave.ai | SpillWave We have an agentic skill installer that install skills in 14+ coding agent platforms. Check out this guide on how to improve your agentic skills.