Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Convert first-time store visitors with behavior-based personalization — browse path → preference segment → matched first-order incentive → popup timing including exit intent. Use when the user wants cold-start offers for new visitors, first-purchase coupons tied to what they viewed, exit-intent or timed modals, or onboarding for anonymous traffic before email capture. Steps baked in: monitor path, classify preference, match first-order voucher or gift, set trigger (exit intent and alternatives).
.claude/skills/leoyeai-new-visitor-cold-start/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 39% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -12% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -25% | 0% |
You turn first session behavior into a single high-fit first-order offer and clear modal timing so new visitors convert without generic blasts.
Read references/cold_start_playbook.md for segment → offer → trigger map.
| Observed path pattern | Preference label | Rationale | |----------------------|------------------|-----------| | e.g. 3 PDPs same category | Category intent | … | | e.g. cart + shipping page | Shipping sensitivity | … |
| Segment | First-order offer | Min order / notes | |---------|-------------------|-------------------| | … | … | … |
| Trigger | When | Copy role | |---------|------|-----------| | Exit intent | Mouse leave viewport / mobile back intent proxy | Save offer + urgency light | | Backup 1 | e.g. 25s on PDP | Deep interest | | Backup 2 | e.g. second collection view | Broad interest |
Include one modal headline + subline + CTA per primary segment (or one adaptive template with dynamic category name).
If user only says "better first-time conversion," still deliver one full row chain: path example → segment → offer → exit-intent line.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 20,577 | 14,135 | -31% | 1 | 1 | 0% | 3,210 | 3,024 | -6% | 0 | 0 | — |
case-02 | fail→fail | 23,937 | 13,039 | -46% | 1 | 1 | 0% | 4,102 | 2,766 | -33% | 0 | 0 | — |
case-03 | fail→fail | 19,264 | 13,086 | -32% | 1 | 1 | 0% | 3,255 | 2,696 | -17% | 0 | 0 | — |
case-04 | fail→pass | 11,466 | 10,624 | -7% | 1 | 1 | 0% | 1,756 | 2,445 | +39% | 0 | 0 | — |
case-05 | fail→pass | 16,538 | 13,718 | -17% | 1 | 1 | 0% | 2,530 | 2,566 | +1% | 0 | 0 | — |
case-06 | pass→pass | 17,884 | 10,224 | -43% | 1 | 1 | 0% | 2,594 | 2,183 | -16% | 0 | 0 | — |
case-07 | pass→pass | 14,358 | 8,203 | -43% | 1 | 1 | 0% | 2,305 | 1,873 | -19% | 0 | 0 | — |
case-08 | fail→pass | 16,466 | 10,313 | -37% | 1 | 1 | 0% | 2,469 | 2,178 | -12% | 0 | 0 | — |
case-09 | fail→pass | 15,140 | 8,333 | -45% | 1 | 1 | 0% | 2,398 | 1,790 | -25% | 0 | 0 | — |
case-10 | fail→fail | 16,420 | 10,648 | -35% | 1 | 1 | 0% | 2,408 | 2,473 | +3% | 0 | 0 | — |
case-11 | pass→pass | 13,358 | 10,995 | -18% | 1 | 1 | 0% | 2,179 | 2,355 | +8% | 0 | 0 | — |
case-12 | fail→pass | 16,866 | 9,014 | -47% | 1 | 1 | 0% | 2,691 | 2,093 | -22% | 0 | 0 | — |
case-13 | pass→pass | 10,415 | 6,680 | -36% | 1 | 1 | 0% | 1,541 | 1,627 | +6% | 0 | 0 | — |
case-14 | pass→pass | 14,220 | 7,632 | -46% | 1 | 1 | 0% | 2,263 | 1,716 | -24% | 0 | 0 | — |
case-15 | pass→pass | 15,251 | 12,516 | -18% | 1 | 1 | 0% | 2,395 | 2,607 | +9% | 0 | 0 | — |
case-16 | fail→pass | 13,307 | 9,204 | -31% | 1 | 1 | 0% | 2,258 | 2,035 | -10% | 0 | 0 | — |
case-17 | fail→pass | 15,983 | 8,146 | -49% | 1 | 1 | 0% | 2,421 | 1,931 | -20% | 0 | 0 | — |
case-18 | pass→pass | 15,469 | 10,165 | -34% | 1 | 1 | 0% | 2,674 | 2,291 | -14% | 0 | 0 | — |
case-19 | pass→pass | 8,956 | 7,005 | -22% | 1 | 1 | 0% | 1,402 | 1,601 | +14% | 0 | 0 | — |
case-20 | pass→pass | 16,790 | 13,757 | -18% | 1 | 1 | 0% | 3,029 | 2,815 | -7% | 0 | 0 | — |
case-21 | pass→pass | 15,077 | 16,143 | +7% | 1 | 1 | 0% | 2,615 | 3,015 | +15% | 0 | 0 | — |
case-22 | pass→fail | 15,463 | 12,723 | -18% | 1 | 1 | 0% | 2,696 | 3,023 | +12% | 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.