Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Generate unique personalized first lines for hundreds of prospects using company news, LinkedIn activity, and mutual connections. Saves 10+ hours of manual research per campaign. Use when you need personalized outreach at volume.
.claude/skills/onewave-ai-personalization-at-scale/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 43% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -4% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -39% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -55% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -53% | 0% |
Generate hundreds of unique, researched first lines in minutes instead of hours, making cold outreach feel warm.
references/research-sources.md - signal sources, personalization styles, quality standardsreferences/patterns-by-type.md - sample first lines and tables for each angle (congrats, observation, mutual connection, company news, hiring, tech stack, thought leadership, shared background)references/fallbacks.md - role/stage/industry/competitor lines for prospects with no anglereferences/output-template.md - full campaign deliverable structurereferences/benchmarks.md - expected lift, A/B reference data, pro tips (do/don't)references/example-campaigns.md - worked campaign examples by personareferences/research-sources.md. Identify the strongest, most recent, verifiable angle per prospect.references/patterns-by-type.md. For prospects with no angle, draft from references/fallbacks.md.references/output-template.md.See references/benchmarks.md for target success rates and references/example-campaigns.md for persona-specific approaches.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 6,486 | 18,797 | +190% | 1 | 1 | 0% | 1,134 | 1,619 | +43% | 0 | 0 | — |
case-02 | fail→pass | 9,118 | 5,767 | -37% | 1 | 1 | 0% | 1,651 | 1,585 | -4% | 0 | 0 | — |
case-03 | fail→fail | 15,478 | 14,937 | -3% | 1 | 1 | 0% | 2,797 | 3,259 | +17% | 0 | 0 | — |
case-04 | pass→pass | 16,180 | 14,788 | -9% | 1 | 1 | 0% | 2,810 | 3,129 | +11% | 0 | 0 | — |
case-05 | pass→pass | 14,922 | 15,872 | +6% | 1 | 1 | 0% | 2,816 | 3,691 | +31% | 0 | 0 | — |
case-06 | pass→pass | 13,124 | 9,771 | -26% | 1 | 1 | 0% | 2,150 | 2,097 | -2% | 0 | 0 | — |
case-07 | fail→fail | 8,031 | 5,363 | -33% | 1 | 1 | 0% | 1,195 | 1,309 | +10% | 0 | 0 | — |
case-08 | pass→pass | 8,111 | 7,236 | -11% | 1 | 1 | 0% | 1,311 | 1,677 | +28% | 0 | 0 | — |
case-09 | fail→pass | 8,306 | 2,436 | -71% | 1 | 1 | 0% | 1,457 | 882 | -39% | 0 | 0 | — |
case-10 | pass→pass | 13,277 | 5,664 | -57% | 1 | 1 | 0% | 2,221 | 1,394 | -37% | 0 | 0 | — |
case-11 | pass→pass | 11,490 | 8,878 | -23% | 1 | 1 | 0% | 1,799 | 2,054 | +14% | 0 | 0 | — |
case-12 | fail→pass | 13,371 | 3,207 | -76% | 1 | 1 | 0% | 2,106 | 956 | -55% | 0 | 0 | — |
case-13 | pass→pass | 8,070 | 2,328 | -71% | 1 | 1 | 0% | 1,302 | 947 | -27% | 0 | 0 | — |
case-14 | fail→pass | 9,196 | 1,853 | -80% | 1 | 1 | 0% | 1,631 | 773 | -53% | 0 | 0 | — |
case-15 | fail→pass | 8,970 | 1,351 | -85% | 1 | 1 | 0% | 1,466 | 703 | -52% | 0 | 0 | — |
case-16 | fail→pass | 12,951 | 6,173 | -52% | 1 | 1 | 0% | 1,853 | 1,585 | -14% | 0 | 0 | — |
case-17 | pass→pass | 11,917 | 8,111 | -32% | 1 | 1 | 0% | 1,916 | 1,216 | -37% | 0 | 0 | — |
case-18 | pass→pass | 6,503 | 4,635 | -29% | 1 | 1 | 0% | 1,165 | 1,275 | +9% | 0 | 0 | — |
case-19 | pass→pass | 8,937 | 7,567 | -15% | 1 | 1 | 0% | 1,492 | 1,583 | +6% | 0 | 0 | — |
case-20 | pass→pass | 10,558 | 5,540 | -48% | 1 | 1 | 0% | 1,807 | 1,402 | -22% | 0 | 0 | — |
case-21 | pass→pass | 9,038 | 3,101 | -66% | 1 | 1 | 0% | 1,410 | 1,047 | -26% | 0 | 0 | — |
case-22 | pass→pass | 11,794 | 12,687 | +8% | 1 | 1 | 0% | 1,923 | 2,259 | +17% | 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.
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.