Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Apply Cialdini's six principles of persuasion — Reciprocity, Commitment/Consistency, Social Proof, Liking, Authority, and Scarcity — to analyze or design influence strategies. Use this skill when the user needs to make messaging more persuasive, analyze why a campaign works or doesn't, design sales or marketing copy, or understand social influence tactics — even if they say 'how do we convince people', 'why is this ad effective', or 'make this more persuasive'.
.claude/skills/asgard-ai-platform-soc-cialdini/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 35% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 10% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 12% | 0% |
Robert Cialdini identified six universal principles that drive human compliance. They work because they're cognitive shortcuts — people rely on them to make quick decisions. Understanding these principles helps design more persuasive communications AND recognize when they're being used on you.
IRON LAW: Ethical Influence, Not Manipulation
These principles are tools. Using them to help people make decisions aligned
with their interests = ethical influence. Using them to exploit people against
their interests = manipulation. The test: would the person feel grateful or
deceived if they knew the technique was being used?1. Reciprocity — People feel obligated to return favors
2. Commitment & Consistency — People want to act consistently with prior commitments
3. Social Proof — People follow what others do, especially similar others
4. Liking — People say yes to those they like
5. Authority — People defer to credible experts
6. Scarcity — People value what's limited or disappearing
markdown# Persuasion Analysis: {Context} ## Current State - Persuader: {who} - Target: {audience} - Desired action: {what} - Current conversion/compliance rate: {if known} ## Principle Audit | Principle | Currently Used? | How | Effectiveness | |-----------|----------------|-----|---------------| | Reciprocity | Y/N | {description} | H/M/L | | Commitment | Y/N | ... | ... | | Social Proof | Y/N | ... | ... | | Liking | Y/N | ... | ... | | Authority | Y/N | ... | ... | | Scarcity | Y/N | ... | ... | ## Recommendations 1. {Principle}: {specific implementation} — {expected impact} ## Ethics Check {Does each recommendation pass the "grateful or deceived?" test?}
Scenario: Improving conversion on a Taiwanese SaaS landing page
| Principle | Current | Recommendation | |-----------|---------|---------------| | Reciprocity | ✗ | Offer a free ROI calculator tool before asking for signup | | Social Proof | Weak ("trusted by companies") | Add specific logos + "2,347 teams use us" + testimonial with photo and company | | Authority | ✗ | Add "Recommended by 資策會" or relevant industry certification | | Scarcity | ✗ | "Early adopter pricing: NT$299/month (ends March 31)" |
All pass ethics check: free tool is genuinely useful, social proof is factual, authority is real certification, scarcity is a genuine time-limited offer ✓
references/dark-patterns.mdreferences/b2b-persuasion.md| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 14,731 | 16,429 | +12% | 1 | 1 | 0% | 2,160 | 3,597 | +67% | 0 | 0 | — |
case-02 | fail→fail | 14,556 | 15,211 | +4% | 1 | 1 | 0% | 2,461 | 3,464 | +41% | 0 | 0 | — |
case-03 | fail→fail | 16,067 | 16,482 | +3% | 1 | 1 | 0% | 2,562 | 3,658 | +43% | 0 | 0 | — |
case-04 | fail→fail | 17,901 | 17,478 | -2% | 1 | 1 | 0% | 2,757 | 3,699 | +34% | 0 | 0 | — |
case-05 | fail→pass | 19,365 | 15,502 | -20% | 1 | 1 | 0% | 2,783 | 3,761 | +35% | 0 | 0 | — |
case-06 | fail→pass | 17,364 | 13,540 | -22% | 1 | 1 | 0% | 2,856 | 3,154 | +10% | 0 | 0 | — |
case-07 | fail→fail | 25,197 | 17,030 | -32% | 1 | 1 | 0% | 3,803 | 3,697 | -3% | 0 | 0 | — |
case-08 | fail→fail | 21,603 | 15,961 | -26% | 1 | 1 | 0% | 3,016 | 3,392 | +12% | 0 | 0 | — |
case-09 | fail→pass | 19,542 | 14,467 | -26% | 1 | 1 | 0% | 2,834 | 3,409 | +20% | 0 | 0 | — |
case-10 | fail→fail | 24,761 | 15,720 | -37% | 1 | 1 | 0% | 3,957 | 3,372 | -15% | 0 | 0 | — |
case-11 | pass→pass | 16,842 | 18,734 | +11% | 1 | 1 | 0% | 3,226 | 4,840 | +50% | 0 | 0 | — |
case-12 | pass→pass | 21,774 | 25,731 | +18% | 1 | 1 | 0% | 4,237 | 6,608 | +56% | 0 | 0 | — |
case-13 | pass→fail | 19,321 | 26,357 | +36% | 1 | 1 | 0% | 2,777 | 5,499 | +98% | 0 | 0 | — |
case-14 | fail→pass | 26,089 | 16,681 | -36% | 1 | 1 | 0% | 3,604 | 3,882 | +8% | 0 | 0 | — |
case-15 | fail→fail | 16,022 | 12,792 | -20% | 1 | 1 | 0% | 2,544 | 3,261 | +28% | 0 | 0 | — |
case-16 | fail→pass | 20,900 | 11,022 | -47% | 1 | 1 | 0% | 2,880 | 3,237 | +12% | 0 | 0 | — |
case-17 | fail→fail | 20,975 | 13,523 | -36% | 1 | 1 | 0% | 2,770 | 3,356 | +21% | 0 | 0 | — |
case-18 | fail→pass | 17,750 | 10,047 | -43% | 1 | 1 | 0% | 2,512 | 2,864 | +14% | 0 | 0 | — |
case-19 | fail→pass | 20,740 | 12,451 | -40% | 1 | 1 | 0% | 3,335 | 3,134 | -6% | 0 | 0 | — |
case-20 | fail→fail | 36,150 | 17,378 | -52% | 1 | 1 | 0% | 4,618 | 3,698 | -20% | 0 | 0 | — |
case-21 | fail→fail | 23,213 | 14,608 | -37% | 1 | 1 | 0% | 3,058 | 3,310 | +8% | 0 | 0 | — |
case-22 | fail→pass | 20,197 | 16,728 | -17% | 1 | 1 | 0% | 2,980 | 3,525 | +18% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +32 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.