Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Plan a paid acquisition / performance marketing program with unit economics that work. Use when asked to plan paid media, allocate an ad budget across channels, set CAC/LTV targets, or structure a creative-testing program. Produces a paid acquisition plan — economic guardrails (CAC/LTV/payback), channel allocation, account & campaign structure, a creative testing plan, the measurement approach, and scale/kill rules.
.claude/skills/mohitagw15856-paid-acquisition-plan/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -27% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -21% | 0% |
| case-17 | ✗→✓ | ▲ Improved | -4% | 0% |
| case-20 | ✗→✓ | ▲ Improved | -16% | 0% |
Paid acquisition is buying customers — it only works if you buy them for less than they're worth, and most plans skip that math. This skill starts from the unit economics (CAC ceiling from LTV and payback), then allocates budget, structures testing, and sets the rules for when to scale a channel and when to kill it.
Ask for these only if they aren't already provided:
1. Economic guardrails — derive the max allowable CAC from LTV × margin ÷ payback target; state the target ROAS and the blended CAC ceiling. Every channel decision flows from this.
2. Channel allocation — a table; weight toward intent and proven channels, reserve a test budget for new ones.
| Channel | Role (intent vs. demand-gen) | Budget % | Target CAC | Why | |---|---|---|---|---|
3. Account & campaign structure — how campaigns/ad sets are organised (by intent, audience, or product), and the budgeting method (e.g. consolidated vs. granular).
4. Creative testing plan — the testing cadence, what varies (hook, format, offer, audience), how many concepts per cycle, and the decision rule for a winner. Creative is the biggest lever in modern paid — treat it as the experiment.
5. Measurement — conversion tracking, the attribution approach and its limits, incrementality testing (geo holdout / lift) for channels that claim credit they didn't earn.
6. Scale & kill rules — the metric thresholds to increase budget on a winner and to cut a loser, and how fast to move (avoid thrashing the learning phase).
Performance-marketing practice — LTV/CAC and payback economics, incrementality testing, and creative-led experimentation.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 25,869 | 24,766 | -4% | 1 | 1 | 0% | 5,150 | 5,315 | +3% | 0 | 0 | — |
case-02 | fail→pass | 32,524 | 21,616 | -34% | 1 | 1 | 0% | 6,266 | 4,572 | -27% | 0 | 0 | — |
case-03 | pass→pass | 29,121 | 30,374 | +4% | 1 | 1 | 0% | 4,834 | 6,308 | +30% | 0 | 0 | — |
case-04 | pass→pass | 22,109 | 24,498 | +11% | 1 | 1 | 0% | 3,675 | 4,590 | +25% | 0 | 0 | — |
case-05 | pass→pass | 14,986 | 15,487 | +3% | 1 | 1 | 0% | 2,546 | 3,338 | +31% | 0 | 0 | — |
case-06 | pass→pass | 18,761 | 18,625 | -1% | 1 | 1 | 0% | 3,118 | 3,962 | +27% | 0 | 0 | — |
case-07 | fail→pass | 15,692 | 11,410 | -27% | 1 | 1 | 0% | 2,803 | 2,865 | +2% | 0 | 0 | — |
case-08 | fail→pass | 17,263 | 10,366 | -40% | 1 | 1 | 0% | 3,165 | 2,485 | -21% | 0 | 0 | — |
case-09 | pass→pass | 16,217 | 13,076 | -19% | 1 | 1 | 0% | 2,525 | 2,719 | +8% | 0 | 0 | — |
case-10 | pass→pass | 15,442 | 11,497 | -26% | 1 | 1 | 0% | 2,360 | 2,440 | +3% | 0 | 0 | — |
case-11 | pass→pass | 11,133 | 15,674 | +41% | 1 | 1 | 0% | 2,136 | 3,190 | +49% | 0 | 0 | — |
case-12 | pass→pass | 17,583 | 14,502 | -18% | 1 | 1 | 0% | 2,855 | 2,858 | +0% | 0 | 0 | — |
case-13 | pass→pass | 15,685 | 11,904 | -24% | 1 | 1 | 0% | 2,454 | 2,615 | +7% | 0 | 0 | — |
case-14 | pass→pass | 18,218 | 15,432 | -15% | 1 | 1 | 0% | 2,974 | 3,141 | +6% | 0 | 0 | — |
case-15 | pass→fail | 16,991 | 14,569 | -14% | 1 | 1 | 0% | 2,713 | 2,911 | +7% | 0 | 0 | — |
case-16 | pass→pass | 14,623 | 12,315 | -16% | 1 | 1 | 0% | 2,418 | 2,668 | +10% | 0 | 0 | — |
case-17 | fail→pass | 13,207 | 8,870 | -33% | 1 | 1 | 0% | 2,248 | 2,164 | -4% | 0 | 0 | — |
case-18 | pass→pass | 17,465 | 13,707 | -22% | 1 | 1 | 0% | 2,875 | 3,114 | +8% | 0 | 0 | — |
case-19 | pass→pass | 10,142 | 5,548 | -45% | 1 | 1 | 0% | 1,749 | 1,581 | -10% | 0 | 0 | — |
case-20 | fail→pass | 15,319 | 9,652 | -37% | 1 | 1 | 0% | 2,997 | 2,527 | -16% | 0 | 0 | — |
case-21 | fail→fail | 19,874 | 14,849 | -25% | 1 | 1 | 0% | 3,664 | 3,040 | -17% | 0 | 0 | — |
case-22 | pass→pass | 17,278 | 10,842 | -37% | 1 | 1 | 0% | 2,657 | 2,289 | -14% | 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 +18 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.