广告创意与文案

好创意降低 50% 获客成本——广告创意是数字广告最大的杠杆。

创意制作流程

文案公式

"""
经典广告文案公式
"""
COPYWRITING_FORMULAS = {
"AIDA": {
"Attention": "吸引注意力 — 问题/数据/挑衅",
"Interest": "引发兴趣 — 痛点共鸣/故事",
"Desire": "激发欲望 — 解决方案/证据",
"Action": "号召行动 — 限时/专属/CTA",
"示例": "还在加班做报表? (A) 每月浪费 40 小时在重复工作 (I) "
"AI 自动报表 3 分钟搞定 (D) 限时免费试用 → (A)",
},
"PAS": {
"Problem": "指出问题",
"Agitate": "放大痛苦",
"Solution": "给出方案",
"示例": "投广告总亏钱? (P) 每天烧掉几百却不知花在哪 (A) "
"我们的 AI 投放让 ROAS 翻倍 (S)",
},
"BAB": {
"Before": "现在的糟糕状态",
"After": "理想的美好状态",
"Bridge": "桥梁 = 你的产品",
"示例": "Before: 团队沟通靠微信群，消息淹没 → "
"After: 任务清晰、进度透明 → Bridge: 试试我们的项目管理工具",
},
"4U": {
"Useful": "有用 — 解决实际问题",
"Urgent": "紧迫 — 限时限量",
"Unique": "独特 — 差异化卖点",
"Ultra-specific": "具体 — 数据和细节",
"示例": "3天内(紧迫) 学会 Python 数据分析(有用), "
"独家项目实战(独特), 已帮助 12,380 人转行(具体)",
},
}
print("=== 广告文案公式 ===")
for formula, details in COPYWRITING_FORMULAS.items():
print(f"\n【{formula} 公式】")
for k, v in details.items():
if k == "示例":
print(f"  📝 {v}")
else:
print(f"  {k}: {v}")

A/B 测试框架

"""
A/B 测试引擎
"""
from dataclasses import dataclass
import math
@dataclass
class ABTest:
"""A/B 测试"""
name: str
variant_a: str
variant_b: str
def evaluate(
self,
a_impressions: int, a_clicks: int, a_conversions: int,
b_impressions: int, b_clicks: int, b_conversions: int,
) -> dict:
a_ctr = a_clicks / a_impressions * 100
b_ctr = b_clicks / b_impressions * 100
a_cvr = a_conversions / a_clicks * 100 if a_clicks else 0
b_cvr = b_conversions / b_clicks * 100 if b_clicks else 0
ctr_lift = (b_ctr - a_ctr) / a_ctr * 100
cvr_lift = (b_cvr - a_cvr) / a_cvr * 100 if a_cvr else 0
# 简化显著性判断
min_sample = 1000
significant = (a_impressions >= min_sample and b_impressions >= min_sample)
winner = "B" if b_cvr > a_cvr else "A"
return {
"测试项": self.name,
"A — CTR": f"{a_ctr:.2f}%",
"B — CTR": f"{b_ctr:.2f}%",
"CTR 提升": f"{ctr_lift:+.1f}%",
"A — CVR": f"{a_cvr:.2f}%",
"B — CVR": f"{b_cvr:.2f}%",
"CVR 提升": f"{cvr_lift:+.1f}%",
"胜出": winner,
"统计显著": "是" if significant else "样本不足",
}
tests = [
ABTest("标题测试", "限时特价", "最后 3 小时! 省 ¥200"),
ABTest("图片测试", "产品白底图", "场景使用图"),
ABTest("CTA 测试", "立即购买", "免费试用 7 天"),
]
test_data = [
(10000, 350, 35, 10000, 520, 52),
(8000, 240, 24, 8000, 320, 38),
(12000, 480, 72, 12000, 600, 108),
]
print("=== A/B 测试结果 ===")
for test, data in zip(tests, test_data):
result = test.evaluate(*data)
print(f"\n{result['测试项']}: {test.variant_a} vs {test.variant_b}")
print(f"  CTR: {result['A — CTR']} → {result['B — CTR']} ({result['CTR 提升']})")
print(f"  CVR: {result['A — CVR']} → {result['B — CVR']} ({result['CVR 提升']})")
print(f"  胜出: 变体 {result['胜出']} | 显著性: {result['统计显著']}")

创意要素清单

创意疲劳管理

小结

• AIDA、PAS、BAB、4U 是四个核心文案公式
• A/B 测试一次只改一个变量，样本量至少 1000
• 创意疲劳信号: CTR 下降 20% 或频次 > 5
• 素材生命周期约 2-4 周，需持续储备新创意

下一章: 追踪与归因