月度投放报告 — 2026 年 7 月
这个 App 的买量从 2026-07-23 开始,所以这里的 7 月只有九天,2026-07-23 至 2026-07-31,前面没有任何 周期可以对比。当时只有 B1 一个广告账户在买量,第二个账户从 2026-08-11 起才开始,在本周期之后。九天 里 B1 在八个广告系列上花了 $2,455.00,拿到 4,561 次安装、单次安装成本 $0.538、注册 5,069 次;入账 收入 $3,125.48,当日回收 $2,344.02,同期群 24 小时口径已经走满,为 $2,697.93。账户自己的投放导 出里,7 月的花费比本报告统计的多,因为账户上还跑过另一个 App 的测试和一笔代投;金额见复现一节。在 账户时区上,我们自己的记录为 Meta 入账口径的 1.07 倍,落在有记录的 0.91–1.06 区间之外。术语定义、 报表时区与三种收入口径见附录。
1 总览
2026-07-23 → 2026-07-31 天数 9 花费 $2,455.00 展示次数 347,058 CPM $7.07 链接点击量 20,124 CTR 5.80% 安装次数 (Meta) 4,561 点击→安装 22.7% CPI $0.538 注册数 (自有) 5,069 入账收入 $3,125.48 (1.27) (Meta 1.18) 当日回收 $2,344.02 (0.95) 同期群 24 小时 $2,697.93 (1.10),覆盖率 100% 当日回收付费用户 233
同期群价值 —— d0 / d3 / d7 的 ARPU 与 ARPPU
按安装日计算,从每个用户自己的安装时刻起算:dN 是他自己的前 (N+1) x 24 小时。ARPU 取自 Adjust,也就是合作方后台看到的那一套;ARPPU 取自我们自己的支付流水,因为 Adjust 没有累计去重付费人数这个字段。窗口还没走完的档位留空,不写 0。
应用 ARPU (d0) ARPU (d3) ARPU (d7) ARPPU (d0) ARPPU (d3) ARPPU (d7) 安装数 Vloom $0.4861 $0.6372 $0.7713 $10.21 $11.35 $12.46 5,524
九天里各广告系列的花费占比: T3(7 月 25 日)55.2%,旧广告系列 14.6%,T2(7 月 25 日)10.9%, T1(7 月 25 日)8.3%,德语区(DE/AT/CH)4.1%,T1(7 月 23 日)2.9%,T2(7 月 23 日)2.3%,T3(7 月 23 日)1.7%。
这个月没有在当天把自己的钱赚回来。 当日回收回报 0.95;入账口径的 1.27 是继承下来的老同期群。
这是本系列里唯一一个不是下限的同期群读数。 H=24 的覆盖率是 100%,$2,697.93 已经冻结。
一个广告系列就是这个月的大部分。 T3(7 月 25 日)在 $2,455.00 里拿走了 $1,355.94,在 347,058 次展示里拿走了 284,604 次。账户级比率就是它自己的比率。
八笔买量之间,单次安装成本跨了两个数量级 ,$0.324 到 $18.098。合并的账户比率读的是当时在跑哪些 测试。
2 分小时,2026-07-23 至 2026-07-31
本节是整个账户,按报表时区,把九天合并在一起看。横轴是一天中的小时,不是时间轴: 每一格里装着九 个小时的买量,每天各出一个,这样做才让一天之内的形状读得出来,也让任何一行都不会被当成一次事件来读 。投放列的分子分母都取自 Meta 的计量口径。付费率、ARPU 与 ARPPU 两侧都是自有口径。两套安装数从不交 叉相除。ours 是该格里我们自己的注册数,cov 是该格同期群在 H=24 上的覆盖率。
小时 花费 展示 CPM 点击 CTR 安装 点击→安装 CPI ours cov 入账 bROAS 当日回收 wROAS wPay 同期群 cROAS 0 $52.96 6,100 8.68 426 6.98% 52 12.2% $1.018 53 100% $38.71 0.73 $0.00 0.00 0 $0.00 0.00 1 $74.99 11,489 6.53 640 5.57% 111 17.3% $0.676 123 100% $88.03 1.17 $52.96 0.71 9 $52.96 0.71 2 $85.60 12,061 7.10 644 5.34% 158 24.5% $0.542 165 100% $40.62 0.47 $68.16 0.80 8 $68.16 0.80 3 $78.74 12,005 6.56 629 5.24% 129 20.5% $0.610 140 100% $156.15 1.98 $166.43 2.11 15 $166.43 2.11 4 $85.43 15,385 5.55 814 5.29% 131 16.1% $0.652 144 100% $107.07 1.25 $90.90 1.06 13 $100.22 1.17 5 $109.12 18,425 5.92 1,003 5.44% 198 19.7% $0.551 218 100% $117.02 1.07 $131.53 1.21 16 $144.25 1.32 6 $116.83 21,582 5.41 1,079 5.00% 251 23.3% $0.465 280 100% $129.15 1.11 $132.99 1.14 13 $136.78 1.17 7 $106.53 18,695 5.70 937 5.01% 219 23.4% $0.486 247 100% $235.86 2.21 $183.85 1.73 21 $203.85 1.91 8 $116.36 21,051 5.53 1,049 4.98% 229 21.8% $0.508 256 100% $181.70 1.56 $170.89 1.47 16 $174.45 1.50 9 $115.87 19,698 5.88 1,050 5.33% 239 22.8% $0.485 272 100% $144.23 1.24 $41.77 0.36 5 $41.77 0.36 10 $119.25 18,397 6.48 1,042 5.66% 236 22.6% $0.505 271 100% $52.02 0.44 $39.47 0.33 5 $46.59 0.39 11 $116.37 16,121 7.22 885 5.49% 213 24.1% $0.546 236 100% $196.42 1.69 $225.41 1.94 15 $289.74 2.49 12 $107.50 13,499 7.96 792 5.87% 184 23.2% $0.584 213 100% $81.37 0.76 $94.78 0.88 10 $113.42 1.06 13 $111.16 14,218 7.82 826 5.81% 193 23.4% $0.576 202 100% $177.24 1.59 $63.93 0.58 9 $63.93 0.58 14 $129.34 14,017 9.23 838 5.98% 190 22.7% $0.681 211 100% $225.72 1.75 $211.67 1.64 12 $229.79 1.78 15 $141.13 17,713 7.97 1,116 6.30% 266 23.8% $0.531 285 100% $169.22 1.20 $146.26 1.04 13 $152.88 1.08 16 $165.90 23,992 6.91 1,373 5.72% 320 23.3% $0.518 365 100% $129.42 0.78 $110.24 0.66 7 $121.00 0.73 17 $149.58 21,681 6.90 1,338 6.17% 321 24.0% $0.466 363 100% $203.44 1.36 $143.32 0.96 16 $174.06 1.16 18 $132.89 18,533 7.17 1,146 6.18% 300 26.2% $0.443 339 100% $129.58 0.98 $80.00 0.60 10 $151.09 1.14 19 $93.47 12,340 7.57 866 7.02% 241 27.8% $0.388 274 100% $126.28 1.35 $61.93 0.66 7 $77.11 0.82 20 $83.63 6,532 12.80 516 7.90% 151 29.3% $0.554 167 100% $122.25 1.46 $59.46 0.71 4 $72.42 0.87 21 $52.16 5,008 10.42 386 7.71% 68 17.6% $0.767 77 100% $108.37 2.08 $25.22 0.48 3 $41.32 0.79 22 $47.69 3,492 13.66 322 9.22% 64 19.9% $0.745 64 100% $83.17 1.74 $32.09 0.67 4 $59.21 1.24 23 $62.50 5,024 12.44 407 8.10% 97 23.8% $0.644 104 100% $82.45 1.32 $10.75 0.17 2 $16.51 0.26 合计 $2,455.00 347,058 7.07 20,124 5.80% 4,561 22.7% $0.538 5,069 100% $3,125.48 1.27 $2,344.02 0.95 233 $2,697.93 1.10
⚠ 每一格的当日回收付费用户数在 0 到 21 之间,只有四格到了 16。合计行看水平,分格只看形状。
每一格的覆盖率都是 100%:这个月已经过去五周才被读。
一天的两头买到的是两种东西。 千次展示成本第 6 小时 5.41、第 22 小时 13.66;点击率的方向正好相反。
最便宜的安装在下午晚些时候,最贵的在午夜前后 :第 19 小时 $0.388,第 0 小时 $1.018;花费的高点 在第 16 小时。
四格过了 16 付费用户门槛,而它们说法不一: 第 5、7、8、17 小时的回报是 1.21、1.73、1.47 与 0.96。
与上一个周期对比
它前面没有周期。买量从 2026-07-23 开始,那正是本报告的第一天,所以逐小时配对、它的符号检验,以及把 上一个周期叠加上来的那张图,都没有对象可比,本月一个都不存在。本节全部是九天的水平读数。本系列能做 的第一次对比,是 8 月对 7 月。
2026-08-16T16:04:18.346126
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
3
6
9
12
15
18
21
hour (UTC)
0
20
40
60
80
100
120
140
160
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Spend by hour
0
3
6
9
12
15
18
21
hour (UTC)
5000
10000
15000
20000
25000
impressions
Impressions and installs by hour
impressions
installs
50
100
150
200
250
300
installs (Meta Leads)
The period's motion — whole account, 2026-07-23..2026-07-31 (UTC)
本周期走势:分小时花费,以及展示与安装对照
2026-08-16T16:04:18.508556
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
4
8
12
16
20
hour (UTC)
5
6
7
8
9
CTR %
CTR
0
4
8
12
16
20
hour (UTC)
6
8
10
12
14
USD per 1,000 impressions
CPM
0
4
8
12
16
20
hour (UTC)
12.5
15.0
17.5
20.0
22.5
25.0
27.5
30.0
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install
0
4
8
12
16
20
hour (UTC)
0.4
0.5
0.6
0.7
0.8
0.9
1.0
USD
Cost per install
Delivery by hour — whole account, 2026-07-23..2026-07-31 (UTC)
全账户分小时的 CTR、CPM、点击→安装与 CPI
2026-08-23T13:56:11.046309
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
IN
MX
US
GB
DE
MY
CH
AU
CA
ID
BR
SA
0
200
400
600
800
1000
1200
USD
delivery raked onto the report clock against two measured margins; revenue is the report period
Spend against within-day revenue by country — 2026-07-01..2026-07-31
spend
within-day revenue
分国家的花费与当日回收对照
2026-08-23T13:56:11.084106
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
−0.04
−0.02
0.00
0.02
0.04
USD per install
Cost per install by country — 2026-07-01..2026-07-31 (report clock)
分国家的单次安装成本
(一)大盘分国家,2026-07-01..2026-07-31。 按报表时区。 Meta 不提供小时 × 国家的交叉表,所以每一份国家切片都要从它自己的账户日折算到 UTC;自有口径本就在 UTC 上。
国家 花费 占比 展示次数 CPM 注册数 (自有) 单次注册成本 当日回收 wROAS 付费用户 IN $1,474.61 60.1% 301,292 $4.89 3,597 $0.410 $1,561.56 1.06 215 MX $236.12 9.6% 16,022 $14.74 295 $0.800 $162.29 0.69 21 US $143.05 5.8% 1,575 $90.80 72 $1.987 $126.80 0.89 8 GB $85.25 3.5% 1,362 $62.59 39 $2.186 $0.00 0.00 0 DE $57.75 2.4% 1,049 $55.04 26 $2.221 $101.66 1.76 3 MY $53.65 2.2% 3,419 $15.69 50 $1.073 $5.38 0.10 1 CH $42.13 1.7% 1,011 $41.69 3 $14.044 $34.65 0.82 1 AU $36.03 1.5% 578 $62.30 15 $2.402 $0.00 0.00 0 CA $32.05 1.3% 669 $47.91 31 $1.034 $19.94 0.62 4 ID $27.23 1.1% 3,947 $6.90 113 $0.241 $4.99 0.18 2 BR $19.73 0.8% 1,369 $14.41 114 $0.173 $5.13 0.26 1 SA $19.65 0.8% 1,073 $18.32 25 $0.786 $0.00 0.00 0
其余 163 个国家未列出 , $227.74 合计(占本表 9.3%)。
上表中没有出现的账户: Nomad Node, KBM1, HR1. 它们的花费在每一张日粒度的表里都有;缺的只是国家切片。
(二)同一切片,每一个账户,2026-07-01..2026-07-31.
B1
国家 花费 占比 展示次数 CPM 注册数 (自有) 单次注册成本 当日回收 wROAS 付费用户 IN $1,474.61 60.1% 301,292 $4.89 3,597 $0.410 $1,561.56 1.06 215 MX $236.12 9.6% 16,022 $14.74 295 $0.800 $162.29 0.69 21 US $143.05 5.8% 1,575 $90.80 72 $1.987 $126.80 0.89 8 GB $85.25 3.5% 1,362 $62.59 39 $2.186 $0.00 0.00 0 DE $57.75 2.4% 1,049 $55.04 26 $2.221 $101.66 1.76 3 MY $53.65 2.2% 3,419 $15.69 50 $1.073 $5.38 0.10 1 CH $42.13 1.7% 1,011 $41.69 3 $14.044 $34.65 0.82 1 AU $36.03 1.5% 578 $62.30 15 $2.402 $0.00 0.00 0 CA $32.05 1.3% 669 $47.91 31 $1.034 $19.94 0.62 4 ID $27.23 1.1% 3,947 $6.90 113 $0.241 $4.99 0.18 2 BR $19.73 0.8% 1,369 $14.41 114 $0.173 $5.13 0.26 1 SA $19.65 0.8% 1,073 $18.32 25 $0.786 $0.00 0.00 0
其余 163 个国家未列出 , $227.74 合计(占本表 9.3%)。
Nomad Node —— 没有分国家的表:它这一天什么都没有投出去,是实测的零。
KBM1 —— 没有分国家的表:它这一天什么都没有投出去,是实测的零。
HR1 —— 没有分国家的表:它这一天什么都没有投出去,是实测的零。
整个区间的对账。 每一个分国家单元格都是一次跨小时的分配;每一个总计都是实测的。每个账户折算出来的国家总计都等于它这些天买到的花费:B1 $2,455.00。逐日的对账——点名每个数字是从哪几个账户日折算来的——在各日报页上。
4 汇总:本周期
本节按 UTC 时钟合并九天。本节平时会带的两样东西都需要第二个周期,而这里没有:一是带双比例检验的合 并逐小时漏斗,二是结构中性的账户行,它要把一个周期的各广告系列数字放到另一个周期的展示结构上重算。 两者都不出现。剩下的是整个月的水平读数,而这一部分本来就不需要对比才成立。
三个收入口径并列
口径 2026-07-23 → 2026-07-31 本期结束后还会动吗 入账收入 $3,125.48 (1.27) (Meta 1.18) 已结算 当日回收 $2,344.02 (0.95) 永远不动 同期群 24 小时 $2,697.93 (1.10),覆盖率 100% 已冻结,覆盖完整
三个口径对这个月的说法差了大约三分之一的收入。 入账口径会算上更早安装的人;同期群 24 小时口 径对一天中的时点中性;当日回收口径动不了。
只作参考、从不排次序:同一批人的迄今同期群收入是 $4,862.31,在 2026-08-16 的付款数据截止时间上 回报 1.98。它说明这批人的钱有多少是在第一天之后才到的,它与任何其他周期都不可比。
入账收入按付费用户已安装的天数拆分
安装于 2026-07-23 → 2026-07-31 当日 $2,344.02 (75.0%) 前 1 日 $349.19 (11.2%) 前 2 日 $285.86 (9.1%) 前 3 日 $74.17 (2.4%) 前 4 日 $45.44 (1.5%) 前 5 日 $21.04 (0.7%) 前 6 日 $5.76 (0.2%)
这张表的第一行就是 当日回收的合计,只是换了一个方向看,这也说明入账与当日回收之间的差额恰好等于继 承下来的老同期群,别无其他。放到一个周期上,这个拆分就不再是钱落在哪一天,而变成了一条衰减曲线。
7 月入账的钱里有四分之三来自当天到达的人,尾巴在三天之内就没了 —— 默认视野取 H=24 依据的正 是这个形状。
2026-08-16T16:04:18.763779
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
booked
within-day
cohort @ 24h
0
500
1000
1500
2000
2500
3000
USD
1.27
0.95
1.10
the number above each bar is that measure's ROAS against the same period's spend
The three revenue measures, 2026-07-23..2026-07-31 (UTC)
2026-07-23..2026-07-31
两个周期的三个收入口径,柱顶标注 ROAS
2026-08-16T16:04:18.806564
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
2026-07-23..2026-07-31
0
500
1000
1500
2000
2500
3000
USD
the bottom segment IS the within-day figure: the same number seen twice
Booked revenue by how long its payer had been installed (UTC)
paid the day they installed
one day after
two days after
three days after
four or more days after
入账收入按付费用户已安装的天数拆分
5 各广告系列并列
投放列的分子分母都取自 Meta 的计量口径,所有收入口径都是自有的。同期群视野 H=24h。报表时区为 UTC 。只按 CPM、单次点击成本与 CPI 排次序 — 这里地域是被有意选择的处理变量,正是这一点让这 几项在广告系列之间可比,而收入仍然需要 16 个付费用户(DECISION_LOG.md #37、#48),未过门槛的广告 系列只能看方向,不能看幅度。
广告系列 花费 占比 展示 展示占比 CPM 点击 CTR 单次点击成本 安装 CPI 当日回收 wROAS 付费用户 T3(7 月 25 日,印度/东南亚) $1,355.94 55.2% 284,604 82.0% 4.76 14,213 4.99% $0.095 2,875 $0.472 $1,316.84 0.97 142 旧广告系列(08-03 停投) $358.20 14.6% 29,592 8.5% 12.10 3,423 11.57% $0.105 1,106 $0.324 $591.31 1.65 61 T2(7 月 25 日) $268.63 10.9% 18,253 5.3% 14.72 1,092 5.98% $0.246 298 $0.901 $144.06 0.54 16 T1(7 月 25 日) $203.93 8.3% 2,591 0.7% 78.71 354 13.66% $0.576 102 $1.999 $87.87 0.43 6 德语区(DE/AT/CH) $99.98 4.1% 2,140 0.6% 46.72 83 3.88% $1.205 23 $4.347 $127.00 1.27 2 T1(7 月 23 日) $72.39 2.9% 1,296 0.4% 55.86 71 5.48% $1.020 4 $18.098 $0.00 0.00 0 T2(7 月 23 日) $55.25 2.3% 2,680 0.8% 20.62 296 11.04% $0.187 35 $1.579 $9.11 0.16 1 T3(7 月 23 日) $40.68 1.7% 5,902 1.7% 6.89 592 10.03% $0.069 118 $0.345 $67.82 1.67 5
表里每一个广告系列都有投放。当一个已经关掉的广告系列在它没有产生的花费上落下转化时,工具会打印一行 "有入账、无投放",这里它一行也没有打印,所以账户合计与八个标签页不用做任何剔除就能对上。
价格按 T1/T2/T3 的次序走,价值反着走 —— 千次展示成本 4.76 到 78.71,单次安装 $0.324 到 $18.098,两者都按分组排序。
三个广告系列到了 16 个付费用户,其中只有一个赚钱: 旧广告系列 1.65,对上 0.97 与 0.54。
旧广告系列是唯一赚回来的一个,而它只跑了九天里的两天。
T1(7 月 23 日)用 $72.39 买了 1,296 次展示,拿到四次安装 ,单次 $18.098。它拿到标签页靠的只是 花费占比。
本节的固定图表组画的是跨天的广告系列花费占比、CPM 与 CPI。图表生成器在这一层不出任何图,所以第 5 节只有表格。
6 广告系列内部
本节是广告层级。先是账户全部素材的清单,按跨广告系列的花费排序;随后每个广告系列一个标签页,在它内 部把同一套表重新切一遍。标签页按花费从高到低排列,每一页都带同样的四节。
一个标签页带什么。 图限定在该广告系列上,表只到素材层;没有任何结论只存在于标签页里。
A. 分素材 — 投放
分子分母都取自 Meta 的计量口径。我们自己的注册数不进入这张表。
素材 花费 展示 CPM 点击 CTR 安装 点击→安装 CPI earner/bodysuit$312.93 26,159 11.96 2,968 11.35% 934 31.47% $0.335 T3b/utility_6s$305.33 48,353 6.31 2,653 5.49% 521 19.64% $0.586 T3b/sasian_spice5_15s$282.88 67,080 4.22 2,184 3.26% 453 20.74% $0.624 T3b/sasian_bodysuit_10s$281.69 54,387 5.18 2,836 5.21% 527 18.58% $0.535 T3b/sasian_bikini_6s$189.22 58,064 3.26 3,145 5.42% 754 23.97% $0.251 T3b/sasian_bikini_10s$120.14 34,303 3.50 1,841 5.37% 332 18.03% $0.362 T1b/utility_6s#2$74.18 877 84.58 148 16.88% 57 38.51% $1.301 T1b/utility_15s$73.98 888 83.31 127 14.30% 35 27.56% $2.114 T2b/utility_6s#3$68.17 3,568 19.11 244 6.84% 72 29.51% $0.947 T2b/utility_15s#2$52.84 3,199 16.52 170 5.31% 37 21.76% $1.428 T2b/latina_es_bikini_10s$42.73 4,706 9.08 291 6.18% 99 34.02% $0.432 T3b/sasian_slip_dress$39.93 5,257 7.60 406 7.72% 81 19.95% $0.493 earner/pool5$36.63 2,737 13.38 372 13.59% 144 38.71% $0.254 T1a/utility_sexy$30.97 559 55.40 24 4.29% 0 0.00% — T1a/control_sexy$28.95 349 82.95 40 11.46% 4 10.00% $7.237 T3a/utility_sexy#2$28.90 4,173 6.93 478 11.45% 114 23.85% $0.254 T2a/control_sexy#2$28.18 1,303 21.63 121 9.29% 13 10.74% $2.168 DE/de_german_mature_6s$27.40 456 60.09 29 6.36% 8 27.59% $3.425 T2a/utility_sexy#3$27.07 1,377 19.66 175 12.71% 22 12.57% $1.230 T3b/sasian_fishnet_top_6s$25.15 4,135 6.08 167 4.04% 46 27.54% $0.547 DE/ch_arab_gulf_6s$21.75 498 43.67 12 2.41% 1 8.33% $21.750 T2b/latina_es_slip_dress$19.39 1,262 15.36 63 4.99% 27 42.86% $0.718 T2b/control_10s$14.50 910 15.93 55 6.04% 3 5.45% $4.833 T3b/sasian_bodysuit_6s$14.50 1,687 8.60 71 4.21% 15 21.13% $0.967 T1b/utility_10s$14.31 232 61.68 16 6.90% 0 0.00% — T1b/control_6s$13.99 188 74.41 19 10.11% 1 5.26% $13.990 T1b/control_15s$13.83 213 64.93 19 8.92% 3 15.79% $4.610 T1b/control_10s#2$13.64 193 70.67 25 12.95% 6 24.00% $2.273 T2b/control_6s#2$13.35 817 16.34 41 5.02% 7 17.07% $1.907 T2b/control_15s#2$13.14 829 15.85 53 6.39% 6 11.32% $2.190 T3b/control_15s#3$13.01 1,483 8.77 124 8.36% 18 14.52% $0.723 T3b/utility_10s#2$12.79 1,287 9.94 138 10.72% 27 19.57% $0.474 T2b/utility_10s#3$12.79 628 20.37 44 7.01% 9 20.45% $1.421 T3b/utility_15s#3$12.66 1,137 11.13 99 8.71% 16 16.16% $0.791 T3b/control_10s#3$12.55 1,267 9.91 105 8.29% 10 9.52% $1.255 T3b/control_6s#3$12.18 1,579 7.71 164 10.39% 32 19.51% $0.381 T3a/control_sexy#3$11.78 1,729 6.81 114 6.59% 4 3.51% $2.945 T3b/sasian_corset_6s$11.62 2,059 5.64 132 6.41% 20 15.15% $0.581 T2b/latina_es_bodysuit_10s$11.01 822 13.39 55 6.69% 13 23.64% $0.847 T3b/sasian_offshoulder$9.32 1,060 8.79 71 6.70% 7 9.86% $1.331 earner/bikini$8.64 696 12.41 83 11.93% 28 33.73% $0.309 T3b/sasian_jean_shorts$6.68 828 8.07 46 5.56% 9 19.57% $0.742 T1a/control_sexy#4$6.56 178 36.85 4 2.25% 0 0.00% — T2b/latina_es_offshoulder$6.04 308 19.61 11 3.57% 3 27.27% $2.013 T1a/utility_sexy#4$5.91 210 28.14 3 1.43% 0 0.00% — DE/de_arab_gulf_10s$5.79 83 69.76 5 6.02% 1 20.00% $5.790 DE/ch_german_10s$5.20 102 50.98 2 1.96% 0 0.00% — DE/ch_latina_true_6s$5.17 106 48.77 1 0.94% 0 0.00% — DE/de_turkish_mature_10s$4.71 126 37.38 5 3.97% 3 60.00% $1.570 T2b/latina_es_bikini_6s$4.15 444 9.35 21 4.73% 7 33.33% $0.593 DE/german_6s$4.07 100 40.70 6 6.00% 0 0.00% — DE/ch_german_mature_6s$4.02 103 39.03 0 0.00% 0 — — DE/turkish_mature$3.95 57 69.30 4 7.02% 4 100.00% $0.987 DE/ch_arab_gulf_10s$3.66 146 25.07 1 0.68% 1 100.00% $3.660 T3b/sasian_fishnet_top_10s$3.63 350 10.37 12 3.43% 3 25.00% $1.210 T2b/latina_es_jean_shorts$3.26 234 13.93 13 5.56% 6 46.15% $0.543 DE/german_mature$3.01 89 33.82 6 6.74% 2 33.33% $1.505 T3b/sasian_corset_10s$2.66 288 9.24 19 6.60% 4 21.05% $0.665 DE/de_arab_gulf_6s$2.51 26 96.54 2 7.69% 1 50.00% $2.510 T2b/latina_es_corset_10s$2.25 192 11.72 12 6.25% 5 41.67% $0.450 T2b/latina_es_bodysuit_6s$2.00 137 14.60 10 7.30% 3 30.00% $0.667 DE/german$1.90 60 31.67 0 0.00% 0 — — T2b/latina_es_spice5_15s$1.46 77 18.96 2 2.60% 1 50.00% $1.460 DE/latina_true$1.36 21 64.76 1 4.76% 0 0.00% — DE/de_turkish_10s$1.16 30 38.67 6 20.00% 2 33.33% $0.580 DE/de_pool5_15s$1.04 21 49.52 0 0.00% 0 — — DE/de_asian_mature_6s$0.99 37 26.76 0 0.00% 0 — — T2b/latina_es_fishnet_top_6s$0.79 75 10.53 3 4.00% 0 0.00% — DE/ch_german_mature_10s$0.76 30 25.33 1 3.33% 0 0.00% — DE/turkish_6s$0.54 24 22.50 1 4.17% 0 0.00% — T2b/latina_es_corset_6s$0.52 7 74.29 2 28.57% 0 0.00% — DE/ch_latina_mature_10s$0.42 7 60.00 0 0.00% 0 — — DE/de_latina_true_6s$0.25 5 50.00 0 0.00% 0 — — T2b/latina_es_fishnet_top_10s$0.24 38 6.32 2 5.26% 0 0.00% — DE/ch_german_6s$0.14 4 35.00 0 0.00% 0 — — DE/ch_asian_mature_6s$0.11 3 36.67 1 33.33% 0 0.00% — DE/ch_pool5_15s$0.07 4 17.50 0 0.00% 0 — —
六条素材拿走了这个月的大部分,其中四条来自南亚素材包。 七十七条素材里有四十六条各自拿到不足 $13。
这张表的上下两半是两个不同的竞价环境。 投印度的素材每千次展示 3.26 到 5.18;T1a/ 与 T1b/ 是 70.67 到 84.58。
2026-08-16T16:04:18.899904
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
2
4
6
8
10
12
14
16
CTR %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CTR by hour — every asset, 2026-07-23..2026-07-31 (UTC)
earner/bodysuit
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T3b/sasian_slip_dress
earner/pool5
T3b/sasian_corset_6s
分素材的分小时 CTR
2026-08-16T16:04:19.057409
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0
10
20
30
40
50
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
earner/bodysuit
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T3b/sasian_slip_dress
earner/pool5
T3b/sasian_corset_6s
分素材的分小时点击→安装
2026-08-16T16:04:19.133383
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.5
1.0
1.5
2.0
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Cost per install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
earner/bodysuit
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T3b/sasian_slip_dress
earner/pool5
T3b/sasian_corset_6s
分素材的分小时单次安装成本
2026-08-16T16:04:18.979701
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
2
4
6
8
10
12
14
16
USD per 1,000 impressions
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CPM by hour — every asset, 2026-07-23..2026-07-31 (UTC)
earner/bodysuit
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T3b/sasian_slip_dress
earner/pool5
T3b/sasian_corset_6s
分素材的分小时 CPM
B. 分素材 — 三种收入口径
每一列都是自有口径。Meta 只能表达 入账,因此它跟在那一列旁边的括号里,永远不单独成行。付费用户是 去重的人,不是交易笔数。同期群视野 H=24h,本周期覆盖率 100% 。报表时区为 UTC 。
素材 入账 bROAS (Meta) 当日回收 wROAS 同期群@H cROAS 付费用户 付费率 ARPU ARPPU 最大付费用户占比 earner/bodysuit$508.49 1.62 (1.71) $508.49 1.62 $603.14 1.93 50 5.0% 0.504 10.17 8.1% T3b/utility_6s$483.71 1.58 (1.11) $226.49 0.74 $314.43 1.03 30 5.1% 0.383 7.55 10.4% T3b/sasian_spice5_15s$300.89 1.06 (0.96) $217.76 0.77 $243.62 0.86 31 5.9% 0.416 7.02 5.9% T3b/sasian_bodysuit_10s$358.25 1.27 (1.35) $280.05 0.99 $325.38 1.16 33 5.4% 0.461 8.49 14.8% T3b/sasian_bikini_6s$401.49 2.12 (1.87) $283.40 1.50 $316.27 1.67 18 2.2% 0.341 15.74 33.1% T3b/sasian_bikini_10s$92.34 0.77 (0.99) $79.46 0.66 $79.46 0.66 7 1.9% 0.215 11.35 52.0% T1b/utility_6s#2$66.93 0.90 (0.90) $66.93 0.90 $66.93 0.90 3 5.2% 1.154 22.31 52.3% T1b/utility_15s$20.94 0.28 (0.28) $20.94 0.28 $20.94 0.28 3 7.9% 0.551 6.98 52.4% T2b/utility_6s#3$27.34 0.40 (0.40) $27.34 0.40 $27.34 0.40 3 4.0% 0.364 9.11 58.2% T2b/utility_15s#2$28.43 0.54 (0.64) $21.63 0.41 $25.03 0.47 2 5.6% 0.601 10.81 73.6% T2b/latina_es_bikini_10s$90.02 2.11 (2.09) $68.40 1.60 $71.80 1.68 9 9.2% 0.698 7.60 33.2% T3b/sasian_slip_dress$95.58 2.39 (1.83) $67.63 1.69 $76.95 1.93 7 7.0% 0.676 9.66 31.0% earner/pool5$77.78 2.12 (2.18) $77.78 2.12 $77.78 2.12 10 6.4% 0.499 7.78 23.1% T1a/utility_sexy$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — T1a/control_sexy$9.39 0.32 (0.32) $0.00 0.00 $9.39 0.32 0 0.0% 0.000 — — T3a/utility_sexy#2$93.57 3.24 (2.23) $67.82 2.35 $71.38 2.47 5 3.7% 0.502 13.56 51.7% T2a/control_sexy#2$9.11 0.32 (0.32) $9.11 0.32 $9.11 0.32 1 7.7% 0.701 9.11 100.0% DE/de_german_mature_6s$101.66 3.71 (3.36) $92.36 3.37 $98.10 3.58 1 12.5% 11.545 92.36 100.0% T2a/utility_sexy#3$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T3b/sasian_fishnet_top_6s$94.54 3.76 (5.02) $88.78 3.53 $94.54 3.76 8 13.6% 1.505 11.10 46.6% DE/ch_arab_gulf_6s$47.02 2.16 (2.15) $34.65 1.59 $47.02 2.16 1 50.0% 17.323 34.65 100.0% T2b/latina_es_slip_dress$5.71 0.29 (0.29) $5.71 0.29 $5.71 0.29 1 3.1% 0.178 5.71 100.0% T2b/control_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T3b/sasian_bodysuit_6s$43.13 2.97 (2.95) $18.63 1.29 $27.95 1.93 2 11.8% 1.096 9.32 69.1% T1b/utility_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — T1b/control_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T1b/control_15s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T1b/control_10s#2$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T2b/control_6s#2$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T2b/control_15s#2$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T3b/control_15s#3$9.32 0.72 (1.25) $9.32 0.72 $9.32 0.72 1 4.8% 0.444 9.32 100.0% T3b/utility_10s#2$38.21 2.99 (0.89) $11.52 0.90 $11.52 0.90 2 6.9% 0.397 5.76 50.0% T2b/utility_10s#3$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T3b/utility_15s#3$10.75 0.85 (1.30) $0.00 0.00 $4.99 0.39 0 0.0% 0.000 — — T3b/control_10s#3$12.88 1.03 (0.00) $3.56 0.28 $3.56 0.28 1 10.0% 0.356 3.56 100.0% T3b/control_6s#3$12.88 1.06 (0.00) $5.76 0.47 $5.76 0.47 1 2.7% 0.156 5.76 100.0% T3a/control_sexy#3$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T3b/sasian_corset_6s$24.50 2.11 (0.79) $24.50 2.11 $24.50 2.11 1 4.5% 1.114 24.50 100.0% T2b/latina_es_bodysuit_10s$9.11 0.83 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T3b/sasian_offshoulder$0.00 0.00 (2.12) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — earner/bikini$5.04 0.58 (0.58) $5.04 0.58 $5.04 0.58 1 3.2% 0.163 5.04 100.0% T3b/sasian_jean_shorts$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T1a/control_sexy#4$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — T2b/latina_es_offshoulder$46.48 7.70 (5.12) $20.99 3.48 $20.99 3.48 1 33.3% 6.997 20.99 100.0% T1a/utility_sexy#4$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/de_arab_gulf_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — DE/ch_german_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/ch_latina_true_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/de_turkish_mature_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T2b/latina_es_bikini_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — DE/german_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/ch_german_mature_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/turkish_mature$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — DE/ch_arab_gulf_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T3b/sasian_fishnet_top_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T2b/latina_es_jean_shorts$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — DE/german_mature$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T3b/sasian_corset_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — DE/de_arab_gulf_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T2b/latina_es_corset_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T2b/latina_es_bodysuit_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — DE/german$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — T2b/latina_es_spice5_15s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — DE/latina_true$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/de_turkish_10s$0.00 0.00 (4.30) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — DE/de_pool5_15s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/de_asian_mature_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — T2b/latina_es_fishnet_top_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/ch_german_mature_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/turkish_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — T2b/latina_es_corset_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/ch_latina_mature_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/de_latina_true_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — T2b/latina_es_fishnet_top_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/ch_german_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/ch_asian_mature_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/ch_pool5_15s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — —
16 付费用户门槛(#37):上表中付费用户数不足 16 的行只能看方向,不能看幅度。
整个月只有五条素材过了门槛,其中四条在同一个广告系列里。 它们的当日回收回报在 0.74 到 1.62 之间。
earner/bodysuit 是本月最不集中的一行收入,也是唯一一个读得出来、又在 1.00 以上的回报 ,最大 的单个付费用户占 8.1%,对上 33.1%。
最大付费用户占比这一列,把本页大多数高回报判了出局 —— 3.37、3.48 与 3.53 各自只站在一到八个 付费用户上。
2026-08-16T16:04:19.700541
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0
2
4
6
8
10
12
14
16
payers / installs %
Payer rate by day — every asset (UTC)
earner/bodysuit
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T1b/utility_6s#2
T1b/utility_15s
T2b/utility_6s#3
T2b/utility_15s#2
T2b/latina_es_bikini_10s
T3b/sasian_slip_dress
earner/pool5
T3a/utility_sexy#2
T2a/utility_sexy#3
T3b/sasian_fishnet_top_6s
T2b/latina_es_slip_dress
T3b/control_15s#3
T3b/utility_10s#2
T3b/control_6s#3
earner/bikini
分素材的每日付费率
2026-08-16T16:04:19.824816
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.00
0.25
0.50
0.75
1.00
1.25
1.50
1.75
within-day USD per install
ARPU by day — every asset (UTC)
earner/bodysuit
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T1b/utility_6s#2
T1b/utility_15s
T2b/utility_6s#3
T2b/utility_15s#2
T2b/latina_es_bikini_10s
T3b/sasian_slip_dress
earner/pool5
T3a/utility_sexy#2
T2a/utility_sexy#3
T3b/sasian_fishnet_top_6s
T2b/latina_es_slip_dress
T3b/control_15s#3
T3b/utility_10s#2
T3b/control_6s#3
earner/bikini
分素材的每日 ARPU
2026-08-16T16:04:19.942814
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0
5
10
15
20
25
within-day USD per payer
ARPPU by day — every asset (UTC)
earner/bodysuit
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T1b/utility_6s#2
T1b/utility_15s
T2b/utility_6s#3
T2b/utility_15s#2
T2b/latina_es_bikini_10s
T3b/sasian_slip_dress
earner/pool5
T3a/utility_sexy#2
T3b/sasian_fishnet_top_6s
T2b/latina_es_slip_dress
T3b/control_15s#3
T3b/utility_10s#2
T3b/control_6s#3
earner/bikini
分素材的每日 ARPPU
2026-08-16T16:04:20.070843
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0
1
2
3
4
revenue / spend
Within-day ROAS by day — every asset (UTC)
earner/bodysuit
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T1b/utility_6s#2
T1b/utility_15s
T2b/utility_6s#3
T2b/utility_15s#2
T2b/latina_es_bikini_10s
T3b/sasian_slip_dress
earner/pool5
T3a/utility_sexy#2
T2a/utility_sexy#3
T3b/sasian_fishnet_top_6s
T2b/latina_es_slip_dress
T3b/control_15s#3
T3b/utility_10s#2
T3b/control_6s#3
earner/bikini
分素材的每日当日回收 ROAS
2026-08-16T16:04:20.224112
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
CTR
click→install
CPI
payer rate *
ARPU *
wROAS *
0.0
0.2
0.4
0.6
0.8
1.0
normalised across the period (1.0 = best of the pack)
* axis sits under the 16-payer bar on most days: direction only, never a level. Assets under 20 installs on the period are not plotted.
Composite — every asset across every measure, 2026-07-23..2026-07-31
earner/bodysuit
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T1b/utility_6s#2
T1b/utility_15s
T2b/utility_6s#3
T2b/utility_15s#2
T2b/latina_es_bikini_10s
T3b/sasian_slip_dress
earner/pool5
T3a/utility_sexy#2
T2a/utility_sexy#3
T3b/sasian_fishnet_top_6s
T2b/latina_es_slip_dress
T3b/utility_10s#2
T3b/control_6s#3
T3b/sasian_corset_6s
earner/bikini
综合图:各素材在各口径上按本周期归一化后的表现
收入类图表中,落在 16 付费用户以下的点画成空心。当日安装不足 20 次的素材从综合图和地域图中剔除。
C. 分素材 — 国家结构
⚠ 本表按账户日切,而本页其余各表都在 UTC 时钟上。 Meta 不提供小时 × 国家的交叉表,因此分国家的投放数据只存 在于广告账户自己的 UTC-7 日上,无法重新拼接。它与上面的 A、B 两表相差七小时;本表只读结构,不与它 们比水平。
B 表里某一行读不读得出来,看这里。最后一列是该素材自身 当日回收中来自美国的占比:当它远高于 该素材的美国花费占比时,这条素材的回报是它买到的那块地域的属性。
素材 花费 印度占比 美国占比 其他占比 美国占其自身收入 earner/bodysuit$408.99 32.8% 4.7% 62.6% 8.6% T3b/utility_6s$305.33 98.9% 0.0% 1.1% 0.0% T3b/sasian_spice5_15s$282.88 100.0% 0.0% 0.0% 0.0% T3b/sasian_bodysuit_10s$281.69 100.0% 0.0% 0.0% 0.0% T3b/sasian_bikini_6s$189.22 100.0% 0.0% 0.0% 0.0% T3b/sasian_bikini_10s$120.14 100.0% 0.0% 0.0% 0.0% T1b/utility_6s#2$74.18 0.0% 48.4% 51.6% 100.0% T1b/utility_15s$73.98 0.0% 42.5% 57.5% 52.4% T2b/utility_6s#3$68.17 0.0% 0.0% 100.0% 0.0% T2b/utility_15s#2$52.84 0.0% 0.0% 100.0% 0.0% T2b/latina_es_bikini_10s$42.73 0.0% 0.0% 100.0% 0.0% T3b/sasian_slip_dress$39.93 100.0% 0.0% 0.0% 0.0% earner/pool5$36.63 27.3% 8.2% 64.5% 6.4% T1a/utility_sexy$30.97 0.0% 51.2% 48.8% — T1a/control_sexy$28.95 0.0% 57.5% 42.5% — T3a/utility_sexy#2$28.90 79.4% 0.0% 20.6% 0.0% T2a/control_sexy#2$28.18 0.0% 0.0% 100.0% 0.0% DE/de_german_mature_6s$27.40 0.0% 0.0% 100.0% 0.0% T2a/utility_sexy#3$27.07 0.0% 0.0% 100.0% — T3b/sasian_fishnet_top_6s$25.15 100.0% 0.0% 0.0% 0.0% DE/ch_arab_gulf_6s$21.75 0.0% 0.0% 100.0% 0.0% T2b/latina_es_slip_dress$19.39 0.0% 0.0% 100.0% 0.0% T2b/control_10s$14.50 0.0% 0.0% 100.0% — T3b/sasian_bodysuit_6s$14.50 100.0% 0.0% 0.0% 0.0% T1b/utility_10s$14.31 0.0% 46.3% 53.7% — T1b/control_6s$13.99 0.0% 39.5% 60.5% — T1b/control_15s$13.83 0.0% 37.4% 62.6% — T1b/control_10s#2$13.64 0.0% 38.8% 61.2% — T2b/control_6s#2$13.35 0.0% 0.0% 100.0% — T2b/control_15s#2$13.14 0.0% 0.0% 100.0% — T3b/control_15s#3$13.01 81.9% 0.0% 18.1% 0.0% T3b/utility_10s#2$12.79 82.5% 0.0% 17.5% 0.0% T2b/utility_10s#3$12.79 0.0% 0.0% 100.0% — T3b/utility_15s#3$12.66 83.9% 0.0% 16.1% — T3b/control_10s#3$12.55 76.6% 0.0% 23.4% 0.0% T3b/control_6s#3$12.18 61.7% 0.0% 38.3% 0.0% T3a/control_sexy#3$11.78 57.8% 0.0% 42.2% — T3b/sasian_corset_6s$11.62 100.0% 0.0% 0.0% 0.0% T2b/latina_es_bodysuit_10s$11.01 0.0% 0.0% 100.0% — T3b/sasian_offshoulder$9.32 100.0% 0.0% 0.0% — earner/bikini$150.33 29.0% 5.0% 65.9% 0.0% T3b/sasian_jean_shorts$6.68 100.0% 0.0% 0.0% — T1a/control_sexy#4$6.56 0.0% 21.2% 78.8% — T2b/latina_es_offshoulder$6.04 0.0% 0.0% 100.0% 0.0% T1a/utility_sexy#4$5.91 0.0% 32.0% 68.0% — DE/de_arab_gulf_10s$5.79 0.0% 0.0% 100.0% — DE/ch_german_10s$5.20 0.0% 0.0% 100.0% — DE/ch_latina_true_6s$5.17 0.0% 0.0% 100.0% — DE/de_turkish_mature_10s$4.71 0.0% 0.0% 100.0% — T2b/latina_es_bikini_6s$4.15 0.0% 0.0% 100.0% — DE/german_6s$4.07 0.0% 0.0% 100.0% — DE/ch_german_mature_6s$4.02 0.0% 0.0% 100.0% — DE/turkish_mature$3.95 0.0% 0.0% 100.0% — DE/ch_arab_gulf_10s$3.66 0.0% 0.0% 100.0% — T3b/sasian_fishnet_top_10s$3.63 100.0% 0.0% 0.0% — T2b/latina_es_jean_shorts$3.26 0.0% 0.0% 100.0% — DE/german_mature$3.01 0.0% 0.0% 100.0% — T3b/sasian_corset_10s$2.66 100.0% 0.0% 0.0% — DE/de_arab_gulf_6s$2.51 0.0% 0.0% 100.0% — T2b/latina_es_corset_10s$2.25 0.0% 0.0% 100.0% — T2b/latina_es_bodysuit_6s$2.00 0.0% 0.0% 100.0% — DE/german$1.90 0.0% 0.0% 100.0% — T2b/latina_es_spice5_15s$1.46 0.0% 0.0% 100.0% — DE/latina_true$1.36 0.0% 0.0% 100.0% — DE/de_turkish_10s$1.16 0.0% 0.0% 100.0% — DE/de_pool5_15s$1.04 0.0% 0.0% 100.0% — DE/de_asian_mature_6s$0.99 0.0% 0.0% 100.0% — T2b/latina_es_fishnet_top_6s$0.79 0.0% 0.0% 100.0% — DE/ch_german_mature_10s$0.76 0.0% 0.0% 100.0% — DE/turkish_6s$0.54 0.0% 0.0% 100.0% — T2b/latina_es_corset_6s$0.52 0.0% 0.0% 100.0% — DE/ch_latina_mature_10s$0.42 0.0% 0.0% 100.0% — DE/de_latina_true_6s$0.25 0.0% 0.0% 100.0% — T2b/latina_es_fishnet_top_10s$0.24 0.0% 0.0% 100.0% — DE/ch_german_6s$0.14 0.0% 0.0% 100.0% — DE/ch_asian_mature_6s$0.11 0.0% 0.0% 100.0% — DE/ch_pool5_15s$0.07 0.0% 0.0% 100.0% —
⚠ 在旧广告系列这几行上,七小时的偏移不是一处小毛病。 本表的窗口是账户自己的日,一直伸进 2026-08-01 UTC,而那时这个广告系列的预算又被调高了。它给出的是 earner/bodysuit $408.99 与 earner/bikini $150.33,A 表给出的是 $312.93 与 $8.64。只读这里的百分比,不要读这里的金额,更不要 拿它去对 A、B 两表。
T1b/utility_6s#2 买到一块地域并在那里赚到了钱,背后只有三个付费用户 —— 48.4% 的钱投在美 国,收入的 100.0% 来自美国。
真正有量的都是印度买量 —— T3b/sasian_ 包 100.0%、T3b/utility_6s 98.9%,里面没有一行是美 国结果。
2026-08-16T16:04:19.297510
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
earner/bodysuit
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
earner/bikini
T3b/sasian_bikini_10s
T1b/utility_6s#2
T1b/utility_15s
T2b/utility_6s#3
T2b/utility_15s#2
T2b/latina_es_bikini_10s
T3b/sasian_slip_dress
earner/pool5
T3a/utility_sexy#2
T2a/utility_sexy#3
T3b/sasian_fishnet_top_6s
T2b/latina_es_slip_dress
T3b/utility_10s#2
T3b/control_6s#3
T3b/sasian_corset_6s
0
20
40
60
80
100
% of that asset's spend
Where each asset bought — share of its own spend by region, 2026-07-23..2026-07-31 (META days)
IN
MX
US
GB
DE
MY
other
各素材自身花费的地域结构
2026-08-16T16:04:19.423341
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
earner/bodysuit
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
earner/bikini
T3b/sasian_bikini_10s
T1b/utility_6s#2
T1b/utility_15s
T2b/utility_6s#3
T2b/utility_15s#2
T2b/latina_es_bikini_10s
T3b/sasian_slip_dress
earner/pool5
T3a/utility_sexy#2
T2a/utility_sexy#3
T3b/sasian_fishnet_top_6s
T2b/latina_es_slip_dress
T3b/utility_10s#2
T3b/control_6s#3
T3b/sasian_corset_6s
0.0
0.5
1.0
1.5
2.0
2.5
3.0
USD per install
Cost per install by region — every asset, 2026-07-23..2026-07-31 (META days)
IN
MX
US
GB
DE
MY
各素材分地域的单次安装成本
2026-08-16T16:04:19.513240
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
earner/bodysuit
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
earner/bikini
T3b/sasian_bikini_10s
T1b/utility_6s#2
T1b/utility_15s
T2b/utility_6s#3
T2b/utility_15s#2
T2b/latina_es_bikini_10s
T3b/sasian_slip_dress
earner/pool5
T3a/utility_sexy#2
T2a/utility_sexy#3
T3b/sasian_fishnet_top_6s
T2b/latina_es_slip_dress
T3b/utility_10s#2
T3b/control_6s#3
T3b/sasian_corset_6s
0
50
100
150
200
250
USD
Within-day revenue by region — every asset, 2026-07-23..2026-07-31 (META days)
IN
MX
US
DE
MY
各素材分地域的当日回收
当前显示 T3(7 月 25 日,印度/东南亚) 旧广告系列(08-03 停投) T2(7 月 25 日) T1(7 月 25 日) 德语区(DE/AT/CH) T1(7 月 23 日) T2(7 月 23 日) T3(7 月 23 日) Show all
T3(7 月 25 日,印度/东南亚) —— 这一行以下的每一节都是 T3(7 月 25 日,印度/东南亚) 的:当日、分小时、分国家、分广告系列、分素材,都记在 T3(7 月 25 日,印度/东南亚) 自己那本账、自己那些账户上。用上面的按钮切换。
2026-07-25 那一轮里投印度与东南亚的一条,也是这个月的主体:账户 $2,455.00 里的 $1,355.94,347,058 次展示里的 284,604 次,2,875 次安装、单次 $0.472,142 个当日回收付费用户,回报 0.97。它是唯一一个 付费用户数明显过了 16 付费用户门槛的广告系列,而即使在它内部,也只有四条素材过得了。
分小时、分国家与汇总
2026-08-16T16:05:15.434676
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
3
6
9
12
15
18
21
hour (UTC)
0
20
40
60
80
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Spend by hour
0
3
6
9
12
15
18
21
hour (UTC)
2500
5000
7500
10000
12500
15000
17500
20000
impressions
Impressions and installs by hour
impressions
installs
0
50
100
150
200
installs (Meta Leads)
The period's motion — whole account, 2026-07-23..2026-07-31 (UTC)
本周期走势:分小时花费,以及展示与安装对照
2026-08-16T16:05:15.607966
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
4
8
12
16
20
hour (UTC)
4.50
4.75
5.00
5.25
5.50
5.75
6.00
CTR %
CTR
0
4
8
12
16
20
hour (UTC)
4.5
5.0
5.5
6.0
USD per 1,000 impressions
CPM
0
4
8
12
16
20
hour (UTC)
5
10
15
20
25
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install
0
4
8
12
16
20
hour (UTC)
0.4
0.6
0.8
1.0
1.2
1.4
1.6
USD
Cost per install
Delivery by hour — T3 (Jul 25, India/SEA), 2026-07-23..2026-07-31 (UTC)
T3 (Jul 25, India/SEA) 的分小时 CTR、CPM、点击→安装与 CPI
2026-08-16T16:05:15.731284
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
IN
ID
PH
PK
NG
0
200
400
600
800
1000
1200
1400
USD
spend and installs are a META-day cut (no hour x country grid exists); revenue is the report period
Spend against within-day revenue by country — 2026-07-23..2026-07-31
spend
within-day revenue
分国家的花费与当日回收对照
2026-08-16T16:05:15.771785
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
IN
ID
PH
PK
0.0
0.1
0.2
0.3
0.4
0.5
0.6
USD per install
Cost per install by country — 2026-07-23..2026-07-31 (META days)
分国家的单次安装成本
2026-08-16T16:05:15.837023
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
booked
within-day
cohort @ 24h
0
250
500
750
1000
1250
1500
1750
2000
USD
1.46
0.97
1.13
the number above each bar is that measure's ROAS against the same period's spend
The three revenue measures, 2026-07-23..2026-07-31 (UTC)
2026-07-23..2026-07-31
两个周期的三个收入口径,柱顶标注 ROAS
2026-08-16T16:05:15.891785
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
2026-07-23..2026-07-31
0
250
500
750
1000
1250
1500
1750
2000
USD
the bottom segment IS the within-day figure: the same number seen twice
Booked revenue by how long its payer had been installed (UTC)
paid the day they installed
one day after
two days after
three days after
four or more days after
入账收入按付费用户已安装的天数拆分
A. 分素材 — 投放
分子分母都取自 Meta 的计量口径。我们自己的注册数不进入这张表。
素材 花费 展示 CPM 点击 CTR 安装 点击→安装 CPI T3b/utility_6s$305.33 48,353 6.31 2,653 5.49% 521 19.64% $0.586 T3b/sasian_spice5_15s$282.88 67,080 4.22 2,184 3.26% 453 20.74% $0.624 T3b/sasian_bodysuit_10s$281.69 54,387 5.18 2,836 5.21% 527 18.58% $0.535 T3b/sasian_bikini_6s$189.22 58,064 3.26 3,145 5.42% 754 23.97% $0.251 T3b/sasian_bikini_10s$120.14 34,303 3.50 1,841 5.37% 332 18.03% $0.362 T3b/sasian_slip_dress$39.93 5,257 7.60 406 7.72% 81 19.95% $0.493 T3b/sasian_fishnet_top_6s$25.15 4,135 6.08 167 4.04% 46 27.54% $0.547 T3b/sasian_bodysuit_6s$14.50 1,687 8.60 71 4.21% 15 21.13% $0.967 T3b/control_15s#3$13.01 1,483 8.77 124 8.36% 18 14.52% $0.723 T3b/utility_10s#2$12.79 1,287 9.94 138 10.72% 27 19.57% $0.474 T3b/utility_15s#3$12.66 1,137 11.13 99 8.71% 16 16.16% $0.791 T3b/control_10s#3$12.55 1,267 9.91 105 8.29% 10 9.52% $1.255 T3b/control_6s#3$12.18 1,579 7.71 164 10.39% 32 19.51% $0.381 T3b/sasian_corset_6s$11.62 2,059 5.64 132 6.41% 20 15.15% $0.581 T3b/sasian_offshoulder$9.32 1,060 8.79 71 6.70% 7 9.86% $1.331 T3b/sasian_jean_shorts$6.68 828 8.07 46 5.56% 9 19.57% $0.742 T3b/sasian_fishnet_top_10s$3.63 350 10.37 12 3.43% 3 25.00% $1.210 T3b/sasian_corset_10s$2.66 288 9.24 19 6.60% 4 21.05% $0.665
五条素材几乎拿走了全部,而这个广告系列内部的成本跨度很大:T3b/sasian_bikini_6s 的千次展示成本是 3.26、单次安装 $0.251,T3b/sasian_offshoulder 是 8.79 与 $1.331。六个 utility_/control_ 格是 07-25 那一轮建的;所有以 sasian_ 命名的,是 2026-07-26 加进一个合并广告组的十二条南亚包素材,所 以这两组之间不构成一次对等的比较。
2026-08-16T16:05:15.971696
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
2
3
4
5
6
7
8
CTR %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CTR by hour — every asset, 2026-07-23..2026-07-31 (UTC)
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T3b/sasian_slip_dress
T3b/sasian_corset_6s
分素材的分小时 CTR
2026-08-16T16:05:16.112715
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0
5
10
15
20
25
30
35
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T3b/sasian_slip_dress
T3b/sasian_corset_6s
分素材的分小时点击→安装
2026-08-16T16:05:16.179466
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.5
1.0
1.5
2.0
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Cost per install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T3b/sasian_slip_dress
T3b/sasian_corset_6s
分素材的分小时单次安装成本
2026-08-16T16:05:16.044907
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
3
4
5
6
7
8
USD per 1,000 impressions
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CPM by hour — every asset, 2026-07-23..2026-07-31 (UTC)
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T3b/sasian_slip_dress
T3b/sasian_corset_6s
分素材的分小时 CPM
B. 分素材 — 三种收入口径
每一列都是自有口径。Meta 只能表达 入账,因此它跟在那一列旁边的括号里,永远不单独成行。付费用户是 去重的人,不是交易笔数。同期群视野 H=24h,本周期覆盖率 100% 。报表时区为 UTC 。
素材 入账 bROAS (Meta) 当日回收 wROAS 同期群@H cROAS 付费用户 付费率 ARPU ARPPU 最大付费用户占比 T3b/utility_6s$483.71 1.58 (1.11) $226.49 0.74 $314.43 1.03 30 5.1% 0.383 7.55 10.4% T3b/sasian_spice5_15s$300.89 1.06 (0.96) $217.76 0.77 $243.62 0.86 31 5.9% 0.416 7.02 5.9% T3b/sasian_bodysuit_10s$358.25 1.27 (1.35) $280.05 0.99 $325.38 1.16 33 5.4% 0.461 8.49 14.8% T3b/sasian_bikini_6s$401.49 2.12 (1.87) $283.40 1.50 $316.27 1.67 18 2.2% 0.341 15.74 33.1% T3b/sasian_bikini_10s$92.34 0.77 (0.99) $79.46 0.66 $79.46 0.66 7 1.9% 0.215 11.35 52.0% T3b/sasian_slip_dress$95.58 2.39 (1.83) $67.63 1.69 $76.95 1.93 7 7.0% 0.676 9.66 31.0% T3b/sasian_fishnet_top_6s$94.54 3.76 (5.02) $88.78 3.53 $94.54 3.76 8 13.6% 1.505 11.10 46.6% T3b/sasian_bodysuit_6s$43.13 2.97 (2.95) $18.63 1.29 $27.95 1.93 2 11.8% 1.096 9.32 69.1% T3b/control_15s#3$9.32 0.72 (1.25) $9.32 0.72 $9.32 0.72 1 4.8% 0.444 9.32 100.0% T3b/utility_10s#2$38.21 2.99 (0.89) $11.52 0.90 $11.52 0.90 2 6.9% 0.397 5.76 50.0% T3b/utility_15s#3$10.75 0.85 (1.30) $0.00 0.00 $4.99 0.39 0 0.0% 0.000 — — T3b/control_10s#3$12.88 1.03 (0.00) $3.56 0.28 $3.56 0.28 1 10.0% 0.356 3.56 100.0% T3b/control_6s#3$12.88 1.06 (0.00) $5.76 0.47 $5.76 0.47 1 2.7% 0.156 5.76 100.0% T3b/sasian_corset_6s$24.50 2.11 (0.79) $24.50 2.11 $24.50 2.11 1 4.5% 1.114 24.50 100.0% T3b/sasian_offshoulder$0.00 0.00 (2.12) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T3b/sasian_jean_shorts$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T3b/sasian_fishnet_top_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T3b/sasian_corset_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — —
16 付费用户门槛(#37):上表中付费用户数不足 16 的行只能看方向,不能看幅度。
四行过了门槛:T3b/sasian_bodysuit_10s 33 个付费用户、0.99,T3b/sasian_spice5_15s 31 个、0.77, T3b/utility_6s 30 个、0.74,T3b/sasian_bikini_6s 18 个、1.50。其中最高的那一行也是最集中的,当 日回收的 33.1% 来自一个人,T3b/sasian_spice5_15s 是 5.9%,所以这几行之间的次序,靠这些数字定 不下来。T3b/sasian_fishnet_top_6s 读出 3.53,背后八个付费用户,该样本量下读不出来。
2026-08-16T16:05:16.551191
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0
2
4
6
8
10
12
14
16
payers / installs %
Payer rate by day — every asset (UTC)
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T3b/sasian_slip_dress
T3b/sasian_fishnet_top_6s
T3b/control_15s#3
T3b/utility_10s#2
T3b/control_6s#3
分素材的每日付费率
2026-08-16T16:05:16.650961
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.00
0.25
0.50
0.75
1.00
1.25
1.50
1.75
within-day USD per install
ARPU by day — every asset (UTC)
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T3b/sasian_slip_dress
T3b/sasian_fishnet_top_6s
T3b/control_15s#3
T3b/utility_10s#2
T3b/control_6s#3
分素材的每日 ARPU
2026-08-16T16:05:16.729755
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0
5
10
15
20
25
within-day USD per payer
ARPPU by day — every asset (UTC)
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T3b/sasian_slip_dress
T3b/sasian_fishnet_top_6s
T3b/control_15s#3
T3b/utility_10s#2
T3b/control_6s#3
分素材的每日 ARPPU
2026-08-16T16:05:16.810507
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0
1
2
3
4
revenue / spend
Within-day ROAS by day — every asset (UTC)
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T3b/sasian_slip_dress
T3b/sasian_fishnet_top_6s
T3b/control_15s#3
T3b/utility_10s#2
T3b/control_6s#3
分素材的每日当日回收 ROAS
2026-08-16T16:05:16.936102
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
CTR
click→install
CPI
payer rate *
ARPU *
wROAS *
0.0
0.2
0.4
0.6
0.8
1.0
normalised across the period (1.0 = best of the pack)
* axis sits under the 16-payer bar on most days: direction only, never a level. Assets under 20 installs on the period are not plotted.
Composite — every asset across every measure, 2026-07-23..2026-07-31
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T3b/sasian_slip_dress
T3b/sasian_fishnet_top_6s
T3b/utility_10s#2
T3b/control_6s#3
T3b/sasian_corset_6s
综合图:各素材在各口径上按本周期归一化后的表现
C. 分素材 — 国家结构
⚠ 本表按账户日切,而本页其余各表都在 UTC 时钟上。 Meta 不提供小时 × 国家的交叉表,因此分国家的投放数据只存 在于广告账户自己的 UTC-7 日上,无法重新拼接。它与上面的 A、B 两表相差七小时;本表只读结构,不与它 们比水平。
B 表里某一行读不读得出来,看这里。最后一列是该素材自身 当日回收中来自美国的占比:当它远高于 该素材的美国花费占比时,这条素材的回报是它买到的那块地域的属性。
素材 花费 印度占比 美国占比 其他占比 美国占其自身收入 T3b/utility_6s$305.33 98.9% 0.0% 1.1% 0.0% T3b/sasian_spice5_15s$282.88 100.0% 0.0% 0.0% 0.0% T3b/sasian_bodysuit_10s$281.69 100.0% 0.0% 0.0% 0.0% T3b/sasian_bikini_6s$189.22 100.0% 0.0% 0.0% 0.0% T3b/sasian_bikini_10s$120.14 100.0% 0.0% 0.0% 0.0% T3b/sasian_slip_dress$39.93 100.0% 0.0% 0.0% 0.0% T3b/sasian_fishnet_top_6s$25.15 100.0% 0.0% 0.0% 0.0% T3b/sasian_bodysuit_6s$14.50 100.0% 0.0% 0.0% 0.0% T3b/control_15s#3$13.01 81.9% 0.0% 18.1% 0.0% T3b/utility_10s#2$12.79 82.5% 0.0% 17.5% 0.0% T3b/utility_15s#3$12.66 83.9% 0.0% 16.1% — T3b/control_10s#3$12.55 76.6% 0.0% 23.4% 0.0% T3b/control_6s#3$12.18 61.7% 0.0% 38.3% 0.0% T3b/sasian_corset_6s$11.62 100.0% 0.0% 0.0% 0.0% T3b/sasian_offshoulder$9.32 100.0% 0.0% 0.0% — T3b/sasian_jean_shorts$6.68 100.0% 0.0% 0.0% — T3b/sasian_fishnet_top_10s$3.63 100.0% 0.0% 0.0% — T3b/sasian_corset_10s$2.66 100.0% 0.0% 0.0% —
sasian_ 包是 100.0% 印度,每一行的美国占比都是 0.0%,所以这里 B 表中没有一行是买地域买来的。五个 control_/utility_ 格的印度占比在 61.7% 到 98.9% 之间,其余落在 T3 的其他国家,这是本表能显示的 唯一一点结构差异,而它很小。
2026-08-16T16:05:16.305100
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T3b/sasian_slip_dress
T3b/sasian_fishnet_top_6s
T3b/utility_10s#2
T3b/control_6s#3
T3b/sasian_corset_6s
0
20
40
60
80
100
% of that asset's spend
Where each asset bought — share of its own spend by region, 2026-07-23..2026-07-31 (META days)
IN
ID
PH
PK
NG
other
各素材自身花费的地域结构
2026-08-16T16:05:16.374089
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T3b/sasian_slip_dress
T3b/sasian_fishnet_top_6s
T3b/utility_10s#2
T3b/control_6s#3
T3b/sasian_corset_6s
0.0
0.2
0.4
0.6
0.8
1.0
1.2
USD per install
Cost per install by region — every asset, 2026-07-23..2026-07-31 (META days)
IN
ID
PH
PK
各素材分地域的单次安装成本
2026-08-16T16:05:16.428482
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
T3b/utility_6s
T3b/sasian_spice5_15s
T3b/sasian_bodysuit_10s
T3b/sasian_bikini_6s
T3b/sasian_bikini_10s
T3b/sasian_slip_dress
T3b/sasian_fishnet_top_6s
T3b/utility_10s#2
T3b/control_6s#3
T3b/sasian_corset_6s
0
50
100
150
200
250
USD
Within-day revenue by region — every asset, 2026-07-23..2026-07-31 (META days)
IN
各素材分地域的当日回收 旧广告系列(08-03 停投) —— 这一行以下的每一节都是 旧广告系列(08-03 停投) 的:当日、分小时、分国家、分广告系列、分素材,都记在 旧广告系列(08-03 停投) 自己那本账、自己那些账户上。用上面的按钮切换。
2026-07-30 开投、2026-08-03 关掉的全球 CBO 买量。在本月它只跑了九天里的两天,花费 $358.20:1,106 次安装、单次 $0.324,是本周期最便宜的;当日回收 $591.31,回报 1.65,背后 61 个付费用户。 DECISION_LOG.md #45 把它定为一条赚钱的投放,一个广告组、CBO、三条素材,所以它的分素材行不是一份 排名。
分小时、分国家与汇总
2026-08-16T16:05:18.803444
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
3
6
9
12
15
18
21
hour (UTC)
0
5
10
15
20
25
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Spend by hour
0
3
6
9
12
15
18
21
hour (UTC)
0
500
1000
1500
2000
2500
3000
impressions
Impressions and installs by hour
impressions
installs
10
20
30
40
50
60
70
80
90
installs (Meta Leads)
The period's motion — whole account, 2026-07-23..2026-07-31 (UTC)
本周期走势:分小时花费,以及展示与安装对照
2026-08-16T16:05:18.971417
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
4
8
12
16
20
hour (UTC)
9
10
11
12
13
14
15
CTR %
CTR
0
4
8
12
16
20
hour (UTC)
8
10
12
14
16
USD per 1,000 impressions
CPM
0
4
8
12
16
20
hour (UTC)
25
30
35
40
45
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install
0
4
8
12
16
20
hour (UTC)
0.250
0.275
0.300
0.325
0.350
0.375
0.400
USD
Cost per install
Delivery by hour — Earner (retired 08-03), 2026-07-23..2026-07-31 (UTC)
Earner (retired 08-03) 的分小时 CTR、CPM、点击→安装与 CPI
2026-08-16T16:05:19.104743
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
IN
BR
US
TR
AR
ID
MX
MY
PH
PE
AU
GB
0
25
50
75
100
125
150
175
USD
spend and installs are a META-day cut (no hour x country grid exists); revenue is the report period
Spend against within-day revenue by country — 2026-07-23..2026-07-31
spend
within-day revenue
分国家的花费与当日回收对照
2026-08-16T16:05:19.155896
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
IN
BR
US
TR
AR
ID
MX
MY
PH
PE
AU
GB
0.0
0.2
0.4
0.6
0.8
1.0
USD per install
Cost per install by country — 2026-07-23..2026-07-31 (META days)
分国家的单次安装成本
2026-08-16T16:05:19.216836
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
booked
within-day
cohort @ 24h
0
100
200
300
400
500
600
700
USD
1.65
1.65
1.92
the number above each bar is that measure's ROAS against the same period's spend
The three revenue measures, 2026-07-23..2026-07-31 (UTC)
2026-07-23..2026-07-31
两个周期的三个收入口径,柱顶标注 ROAS
2026-08-16T16:05:19.267444
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
2026-07-23..2026-07-31
0
100
200
300
400
500
USD
the bottom segment IS the within-day figure: the same number seen twice
Booked revenue by how long its payer had been installed (UTC)
paid the day they installed
one day after
two days after
three days after
four or more days after
入账收入按付费用户已安装的天数拆分
A. 分素材 — 投放
分子分母都取自 Meta 的计量口径。我们自己的注册数不进入这张表。
素材 花费 展示 CPM 点击 CTR 安装 点击→安装 CPI earner/bodysuit$312.93 26,159 11.96 2,968 11.35% 934 31.47% $0.335 earner/pool5$36.63 2,737 13.38 372 13.59% 144 38.71% $0.254 earner/bikini$8.64 696 12.41 83 11.93% 28 33.73% $0.309
CBO 把这个广告系列 $358.20 里的 $312.93 放在 earner/bodysuit 上,earner/bikini 只拿到 $8.64。 这正是本格式在广告组层级提醒过的集中:花费这么不均的一个广告组,是很早就挑定了一条素材,三条从来没 有排出过次序。三条在每一项投放比率上都很接近,CPM 是 11.96、13.38 与 12.41,单次安装成本是 $0.335 、$0.254 与 $0.309,所以这个分配从表上看不出任何依据。
2026-08-16T16:05:19.330952
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
9
10
11
12
13
14
15
CTR %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CTR by hour — every asset, 2026-07-23..2026-07-31 (UTC)
earner/bodysuit
earner/pool5
分素材的分小时 CTR
2026-08-16T16:05:19.436126
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
25
30
35
40
45
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
earner/bodysuit
earner/pool5
分素材的分小时点击→安装
2026-08-16T16:05:19.490250
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.225
0.250
0.275
0.300
0.325
0.350
0.375
0.400
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Cost per install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
earner/bodysuit
earner/pool5
分素材的分小时单次安装成本
2026-08-16T16:05:19.382198
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
8
10
12
14
16
USD per 1,000 impressions
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CPM by hour — every asset, 2026-07-23..2026-07-31 (UTC)
earner/bodysuit
earner/pool5
分素材的分小时 CPM
B. 分素材 — 三种收入口径
每一列都是自有口径。Meta 只能表达 入账,因此它跟在那一列旁边的括号里,永远不单独成行。付费用户是 去重的人,不是交易笔数。同期群视野 H=24h,本周期覆盖率 100% 。报表时区为 UTC 。
素材 入账 bROAS (Meta) 当日回收 wROAS 同期群@H cROAS 付费用户 付费率 ARPU ARPPU 最大付费用户占比 earner/bodysuit$508.49 1.62 (1.71) $508.49 1.62 $603.14 1.93 50 5.0% 0.504 10.17 8.1% earner/pool5$77.78 2.12 (2.18) $77.78 2.12 $77.78 2.12 10 6.4% 0.499 7.78 23.1% earner/bikini$5.04 0.58 (0.58) $5.04 0.58 $5.04 0.58 1 3.2% 0.163 5.04 100.0%
16 付费用户门槛(#37):上表中付费用户数不足 16 的行只能看方向,不能看幅度。
只有 earner/bodysuit 过了门槛,50 个付费用户,最大的单个付费用户占其当日回收的 8.1%。这两点 加在一起,让它的 1.62 成为本月唯一一个可以当成水平来读的分素材回报。earner/pool5 读出 2.12,背后 十个付费用户;earner/bikini 读出 0.58,背后一个人;两者在该样本量下都读不出来。它们跑的那两天里 ,三条素材下面的预算从每日 $150 一路动到 $250、$500 与 $875,所以这里也没有任何一处是同预算区间的 。
2026-08-16T16:05:19.797499
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0
1
2
3
4
5
6
7
8
payers / installs %
Payer rate by day — every asset (UTC)
earner/bodysuit
earner/pool5
earner/bikini
分素材的每日付费率
2026-08-16T16:05:19.851812
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
within-day USD per install
ARPU by day — every asset (UTC)
earner/bodysuit
earner/pool5
earner/bikini
分素材的每日 ARPU
2026-08-16T16:05:19.905335
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0
2
4
6
8
10
within-day USD per payer
ARPPU by day — every asset (UTC)
earner/bodysuit
earner/pool5
earner/bikini
分素材的每日 ARPPU
2026-08-16T16:05:19.971070
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.5
1.0
1.5
2.0
2.5
3.0
revenue / spend
Within-day ROAS by day — every asset (UTC)
earner/bodysuit
earner/pool5
earner/bikini
分素材的每日当日回收 ROAS
2026-08-16T16:05:20.050742
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
CTR
click→install
CPI
payer rate *
ARPU *
wROAS *
0.0
0.2
0.4
0.6
0.8
1.0
normalised across the period (1.0 = best of the pack)
* axis sits under the 16-payer bar on most days: direction only, never a level. Assets under 20 installs on the period are not plotted.
Composite — every asset across every measure, 2026-07-23..2026-07-31
earner/bodysuit
earner/pool5
earner/bikini
综合图:各素材在各口径上按本周期归一化后的表现
C. 分素材 — 国家结构
⚠ 本表按账户日切,而本页其余各表都在 UTC 时钟上。 Meta 不提供小时 × 国家的交叉表,因此分国家的投放数据只存 在于广告账户自己的 UTC-7 日上,无法重新拼接。它与上面的 A、B 两表相差七小时;本表只读结构,不与它 们比水平。
B 表里某一行读不读得出来,看这里。最后一列是该素材自身 当日回收中来自美国的占比:当它远高于 该素材的美国花费占比时,这条素材的回报是它买到的那块地域的属性。
素材 花费 印度占比 美国占比 其他占比 美国占其自身收入 earner/bodysuit$408.99 32.8% 4.7% 62.6% 8.6% earner/pool5$36.63 27.3% 8.2% 64.5% 6.4% earner/bikini$150.33 29.0% 5.0% 65.9% 0.0%
⚠ 时钟偏移在这里造成的是实打实的偏差。 账户自己的日比报告的日晚七小时,一直伸进 2026-08-01 UTC ,而那时这个广告系列的预算又被调高了。这里的花费列读出 $408.99 与 $150.33,A 表读出的是 $312.93 与 $8.64,所以这些金额属于另一个窗口,只有百分比能对得过来。
结构本身就是一次不设限的全球买量要看的东西:三条素材都是大约三分之一印度、二十分之一美国,其余分散 。earner/bodysuit 自身收入里的美国占比是 8.6%,它的美国花费占比是 4.7%,这是本页上这个差距最温和 的一次。
2026-08-16T16:05:19.599159
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
earner/bodysuit
earner/bikini
earner/pool5
0
20
40
60
80
100
% of that asset's spend
Where each asset bought — share of its own spend by region, 2026-07-23..2026-07-31 (META days)
IN
BR
US
TR
AR
ID
other
各素材自身花费的地域结构
2026-08-16T16:05:19.649613
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
earner/bodysuit
earner/bikini
earner/pool5
0.0
0.2
0.4
0.6
0.8
USD per install
Cost per install by region — every asset, 2026-07-23..2026-07-31 (META days)
IN
BR
US
TR
AR
ID
各素材分地域的单次安装成本
2026-08-16T16:05:19.699085
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
earner/bodysuit
earner/bikini
earner/pool5
0
20
40
60
80
100
120
140
160
USD
Within-day revenue by region — every asset, 2026-07-23..2026-07-31 (META days)
IN
BR
US
TR
AR
ID
各素材分地域的当日回收 T2(7 月 25 日) —— 这一行以下的每一节都是 T2(7 月 25 日) 的:当日、分小时、分国家、分广告系列、分素材,都记在 T2(7 月 25 日) 自己那本账、自己那些账户上。用上面的按钮切换。
07-25 那一轮的中间一档,买阿联酋、沙特、马来西亚、南非与墨西哥,另外还有 2026-07-26 加进来的西语拉 丁裔素材包。$268.63 买了 18,253 次展示,298 次安装、单次 $0.901,当日回收 $144.06、回报 0.54, 背后正好 16 个付费用户,刚好卡在门槛上。
分小时、分国家与汇总
2026-08-16T16:05:21.942672
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
3
6
9
12
15
18
21
hour (UTC)
0
5
10
15
20
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Spend by hour
0
3
6
9
12
15
18
21
hour (UTC)
200
400
600
800
1000
1200
1400
1600
impressions
Impressions and installs by hour
impressions
installs
0
5
10
15
20
25
installs (Meta Leads)
The period's motion — whole account, 2026-07-23..2026-07-31 (UTC)
本周期走势:分小时花费,以及展示与安装对照
2026-08-16T16:05:22.104345
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
4
8
12
16
20
hour (UTC)
4
5
6
7
8
CTR %
CTR
0
4
8
12
16
20
hour (UTC)
12
13
14
15
16
17
18
19
USD per 1,000 impressions
CPM
0
4
8
12
16
20
hour (UTC)
0
10
20
30
40
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install
0
4
8
12
16
20
hour (UTC)
0.6
0.8
1.0
1.2
1.4
1.6
USD
Cost per install
Delivery by hour — T2 (Jul 25), 2026-07-23..2026-07-31 (UTC)
T2 (Jul 25) 的分小时 CTR、CPM、点击→安装与 CPI
2026-08-16T16:05:22.226177
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
MX
MY
SA
AE
ZA
0
25
50
75
100
125
150
175
200
USD
spend and installs are a META-day cut (no hour x country grid exists); revenue is the report period
Spend against within-day revenue by country — 2026-07-23..2026-07-31
spend
within-day revenue
分国家的花费与当日回收对照
2026-08-16T16:05:22.267280
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
MX
MY
SA
AE
ZA
0.0
0.5
1.0
1.5
2.0
USD per install
Cost per install by country — 2026-07-23..2026-07-31 (META days)
分国家的单次安装成本
2026-08-16T16:05:22.327391
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
booked
within-day
cohort @ 24h
0
25
50
75
100
125
150
175
200
USD
0.77
0.54
0.56
the number above each bar is that measure's ROAS against the same period's spend
The three revenue measures, 2026-07-23..2026-07-31 (UTC)
2026-07-23..2026-07-31
两个周期的三个收入口径,柱顶标注 ROAS
2026-08-16T16:05:22.374436
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
2026-07-23..2026-07-31
0
25
50
75
100
125
150
175
200
USD
the bottom segment IS the within-day figure: the same number seen twice
Booked revenue by how long its payer had been installed (UTC)
paid the day they installed
one day after
two days after
three days after
four or more days after
入账收入按付费用户已安装的天数拆分
A. 分素材 — 投放
分子分母都取自 Meta 的计量口径。我们自己的注册数不进入这张表。
素材 花费 展示 CPM 点击 CTR 安装 点击→安装 CPI T2b/utility_6s#3$68.17 3,568 19.11 244 6.84% 72 29.51% $0.947 T2b/utility_15s#2$52.84 3,199 16.52 170 5.31% 37 21.76% $1.428 T2b/latina_es_bikini_10s$42.73 4,706 9.08 291 6.18% 99 34.02% $0.432 T2b/latina_es_slip_dress$19.39 1,262 15.36 63 4.99% 27 42.86% $0.718 T2b/control_10s$14.50 910 15.93 55 6.04% 3 5.45% $4.833 T2b/control_6s#2$13.35 817 16.34 41 5.02% 7 17.07% $1.907 T2b/control_15s#2$13.14 829 15.85 53 6.39% 6 11.32% $2.190 T2b/utility_10s#3$12.79 628 20.37 44 7.01% 9 20.45% $1.421 T2b/latina_es_bodysuit_10s$11.01 822 13.39 55 6.69% 13 23.64% $0.847 T2b/latina_es_offshoulder$6.04 308 19.61 11 3.57% 3 27.27% $2.013 T2b/latina_es_bikini_6s$4.15 444 9.35 21 4.73% 7 33.33% $0.593 T2b/latina_es_jean_shorts$3.26 234 13.93 13 5.56% 6 46.15% $0.543 T2b/latina_es_corset_10s$2.25 192 11.72 12 6.25% 5 41.67% $0.450 T2b/latina_es_bodysuit_6s$2.00 137 14.60 10 7.30% 3 30.00% $0.667 T2b/latina_es_spice5_15s$1.46 77 18.96 2 2.60% 1 50.00% $1.460 T2b/latina_es_fishnet_top_6s$0.79 75 10.53 3 4.00% 0 0.00% — T2b/latina_es_corset_6s$0.52 7 74.29 2 28.57% 0 0.00% — T2b/latina_es_fishnet_top_10s$0.24 38 6.32 2 5.26% 0 0.00% —
加进来的拉丁裔素材包,比它所在的那一轮买得更好 —— T2b/latina_es_bikini_10s 9.08 与 $0.432,T2b/utility_6s#3 19.11 与 $0.947。
2026-08-16T16:05:22.439755
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
CTR %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CTR by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时 CTR
2026-08-16T16:05:22.538289
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时点击→安装
2026-08-16T16:05:22.588017
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Cost per install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时单次安装成本
2026-08-16T16:05:22.487696
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
USD per 1,000 impressions
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CPM by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时 CPM
B. 分素材 — 三种收入口径
每一列都是自有口径。Meta 只能表达 入账,因此它跟在那一列旁边的括号里,永远不单独成行。付费用户是 去重的人,不是交易笔数。同期群视野 H=24h,本周期覆盖率 100% 。报表时区为 UTC 。
素材 入账 bROAS (Meta) 当日回收 wROAS 同期群@H cROAS 付费用户 付费率 ARPU ARPPU 最大付费用户占比 T2b/utility_6s#3$27.34 0.40 (0.40) $27.34 0.40 $27.34 0.40 3 4.0% 0.364 9.11 58.2% T2b/utility_15s#2$28.43 0.54 (0.64) $21.63 0.41 $25.03 0.47 2 5.6% 0.601 10.81 73.6% T2b/latina_es_bikini_10s$90.02 2.11 (2.09) $68.40 1.60 $71.80 1.68 9 9.2% 0.698 7.60 33.2% T2b/latina_es_slip_dress$5.71 0.29 (0.29) $5.71 0.29 $5.71 0.29 1 3.1% 0.178 5.71 100.0% T2b/control_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T2b/control_6s#2$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T2b/control_15s#2$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T2b/utility_10s#3$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T2b/latina_es_bodysuit_10s$9.11 0.83 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T2b/latina_es_offshoulder$46.48 7.70 (5.12) $20.99 3.48 $20.99 3.48 1 33.3% 6.997 20.99 100.0% T2b/latina_es_bikini_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T2b/latina_es_jean_shorts$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T2b/latina_es_corset_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T2b/latina_es_bodysuit_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T2b/latina_es_spice5_15s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T2b/latina_es_fishnet_top_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — T2b/latina_es_corset_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — T2b/latina_es_fishnet_top_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — —
16 付费用户门槛(#37):上表中付费用户数不足 16 的行只能看方向,不能看幅度。
这里没有一条素材过得了门槛。这个广告系列的 16 个付费用户分散在五条素材上,最多的一条占九个,十八条 素材里有十二条在任何口径上都没有收入。两个看起来像结果的数字,T2b/latina_es_offshoulder 的 3.48 与 T2b/latina_es_bikini_10s 的 1.60,一个建立在一个付费用户上,另一个建立在九个上、其中一个人占 了三分之一。该样本量下读不出来。
2026-08-16T16:05:22.877429
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0
2
4
6
8
10
payers / installs %
Payer rate by day — every asset (UTC)
T2b/utility_6s#3
T2b/utility_15s#2
T2b/latina_es_bikini_10s
T2b/latina_es_slip_dress
分素材的每日付费率
2026-08-16T16:05:22.931663
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
within-day USD per install
ARPU by day — every asset (UTC)
T2b/utility_6s#3
T2b/utility_15s#2
T2b/latina_es_bikini_10s
T2b/latina_es_slip_dress
分素材的每日 ARPU
2026-08-16T16:05:22.987148
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0
2
4
6
8
10
within-day USD per payer
ARPPU by day — every asset (UTC)
T2b/utility_6s#3
T2b/utility_15s#2
T2b/latina_es_bikini_10s
T2b/latina_es_slip_dress
分素材的每日 ARPPU
2026-08-16T16:05:23.047184
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
revenue / spend
Within-day ROAS by day — every asset (UTC)
T2b/utility_6s#3
T2b/utility_15s#2
T2b/latina_es_bikini_10s
T2b/latina_es_slip_dress
分素材的每日当日回收 ROAS
2026-08-16T16:05:23.144719
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
CTR
click→install
CPI
payer rate *
ARPU *
wROAS *
0.0
0.2
0.4
0.6
0.8
1.0
normalised across the period (1.0 = best of the pack)
* axis sits under the 16-payer bar on most days: direction only, never a level. Assets under 20 installs on the period are not plotted.
Composite — every asset across every measure, 2026-07-23..2026-07-31
T2b/utility_6s#3
T2b/utility_15s#2
T2b/latina_es_bikini_10s
T2b/latina_es_slip_dress
综合图:各素材在各口径上按本周期归一化后的表现
C. 分素材 — 国家结构
⚠ 本表按账户日切,而本页其余各表都在 UTC 时钟上。 Meta 不提供小时 × 国家的交叉表,因此分国家的投放数据只存 在于广告账户自己的 UTC-7 日上,无法重新拼接。它与上面的 A、B 两表相差七小时;本表只读结构,不与它 们比水平。
B 表里某一行读不读得出来,看这里。最后一列是该素材自身 当日回收中来自美国的占比:当它远高于 该素材的美国花费占比时,这条素材的回报是它买到的那块地域的属性。
素材 花费 印度占比 美国占比 其他占比 美国占其自身收入 T2b/utility_6s#3$68.17 0.0% 0.0% 100.0% 0.0% T2b/utility_15s#2$52.84 0.0% 0.0% 100.0% 0.0% T2b/latina_es_bikini_10s$42.73 0.0% 0.0% 100.0% 0.0% T2b/latina_es_slip_dress$19.39 0.0% 0.0% 100.0% 0.0% T2b/control_10s$14.50 0.0% 0.0% 100.0% — T2b/control_6s#2$13.35 0.0% 0.0% 100.0% — T2b/control_15s#2$13.14 0.0% 0.0% 100.0% — T2b/utility_10s#3$12.79 0.0% 0.0% 100.0% — T2b/latina_es_bodysuit_10s$11.01 0.0% 0.0% 100.0% — T2b/latina_es_offshoulder$6.04 0.0% 0.0% 100.0% 0.0% T2b/latina_es_bikini_6s$4.15 0.0% 0.0% 100.0% — T2b/latina_es_jean_shorts$3.26 0.0% 0.0% 100.0% — T2b/latina_es_corset_10s$2.25 0.0% 0.0% 100.0% — T2b/latina_es_bodysuit_6s$2.00 0.0% 0.0% 100.0% — T2b/latina_es_spice5_15s$1.46 0.0% 0.0% 100.0% — T2b/latina_es_fishnet_top_6s$0.79 0.0% 0.0% 100.0% — T2b/latina_es_corset_6s$0.52 0.0% 0.0% 100.0% — T2b/latina_es_fishnet_top_10s$0.24 0.0% 0.0% 100.0% —
每一行都是 0.0% 印度、0.0% 美国、100.0% 其他,这正是这个广告系列的地域设置在起作用。这一列分不开墨 西哥与沙特,所以这里的结构问题由第 3 节回答:墨西哥 $243.10、回报 0.67,马来西亚 $59.46、回报 0.09 。
2026-08-16T16:05:22.684185
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
T2b/utility_6s#3
T2b/utility_15s#2
T2b/latina_es_bikini_10s
T2b/latina_es_slip_dress
0
20
40
60
80
100
% of that asset's spend
Where each asset bought — share of its own spend by region, 2026-07-23..2026-07-31 (META days)
MX
MY
SA
AE
ZA
other
各素材自身花费的地域结构
2026-08-16T16:05:22.735823
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
T2b/utility_6s#3
T2b/utility_15s#2
T2b/latina_es_bikini_10s
T2b/latina_es_slip_dress
0.0
0.5
1.0
1.5
2.0
2.5
3.0
USD per install
Cost per install by region — every asset, 2026-07-23..2026-07-31 (META days)
MX
MY
SA
AE
ZA
各素材分地域的单次安装成本
2026-08-16T16:05:22.786891
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
T2b/utility_6s#3
T2b/utility_15s#2
T2b/latina_es_bikini_10s
T2b/latina_es_slip_dress
0
10
20
30
40
50
60
70
USD
Within-day revenue by region — every asset, 2026-07-23..2026-07-31 (META days)
MX
各素材分地域的当日回收 T1(7 月 25 日) —— 这一行以下的每一节都是 T1(7 月 25 日) 的:当日、分小时、分国家、分广告系列、分素材,都记在 T1(7 月 25 日) 自己那本账、自己那些账户上。用上面的按钮切换。
07-25 那一轮的英语一档:美国、加拿大、英国、澳大利亚与新西兰。$203.93 买了 2,591 次展示、CPM 78.71 ,102 次安装、单次 $1.999;当日回收 $87.87、回报 0.43,背后六个付费用户。
分小时、分国家与汇总
2026-08-16T16:05:25.054190
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
3
6
9
12
15
18
21
hour (UTC)
0
2
4
6
8
10
12
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Spend by hour
0
3
6
9
12
15
18
21
hour (UTC)
40
60
80
100
120
140
160
180
impressions
Impressions and installs by hour
impressions
installs
0
2
4
6
8
10
installs (Meta Leads)
The period's motion — whole account, 2026-07-23..2026-07-31 (UTC)
本周期走势:分小时花费,以及展示与安装对照
2026-08-16T16:05:25.203202
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
4
8
12
16
20
hour (UTC)
−0.04
−0.02
0.00
0.02
0.04
CTR %
CTR
0
4
8
12
16
20
hour (UTC)
−0.04
−0.02
0.00
0.02
0.04
USD per 1,000 impressions
CPM
0
4
8
12
16
20
hour (UTC)
−0.04
−0.02
0.00
0.02
0.04
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install
0
4
8
12
16
20
hour (UTC)
−0.04
−0.02
0.00
0.02
0.04
USD
Cost per install
Delivery by hour — T1 (Jul 25), 2026-07-23..2026-07-31 (UTC)
T1 (Jul 25) 的分小时 CTR、CPM、点击→安装与 CPI
2026-08-16T16:05:25.326782
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
US
GB
AU
CA
NZ
VI
0
20
40
60
80
USD
spend and installs are a META-day cut (no hour x country grid exists); revenue is the report period
Spend against within-day revenue by country — 2026-07-23..2026-07-31
spend
within-day revenue
分国家的花费与当日回收对照
2026-08-16T16:05:25.366829
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
US
GB
AU
CA
NZ
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
USD per install
Cost per install by country — 2026-07-23..2026-07-31 (META days)
分国家的单次安装成本
2026-08-16T16:05:25.423860
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
booked
within-day
cohort @ 24h
0
20
40
60
80
USD
0.43
0.43
0.43
the number above each bar is that measure's ROAS against the same period's spend
The three revenue measures, 2026-07-23..2026-07-31 (UTC)
2026-07-23..2026-07-31
两个周期的三个收入口径,柱顶标注 ROAS
2026-08-16T16:05:25.467753
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
2026-07-23..2026-07-31
0
10
20
30
40
50
60
70
80
USD
the bottom segment IS the within-day figure: the same number seen twice
Booked revenue by how long its payer had been installed (UTC)
paid the day they installed
one day after
two days after
three days after
four or more days after
入账收入按付费用户已安装的天数拆分
A. 分素材 — 投放
分子分母都取自 Meta 的计量口径。我们自己的注册数不进入这张表。
素材 花费 展示 CPM 点击 CTR 安装 点击→安装 CPI T1b/utility_6s#2$74.18 877 84.58 148 16.88% 57 38.51% $1.301 T1b/utility_15s$73.98 888 83.31 127 14.30% 35 27.56% $2.114 T1b/utility_10s$14.31 232 61.68 16 6.90% 0 0.00% — T1b/control_6s$13.99 188 74.41 19 10.11% 1 5.26% $13.990 T1b/control_15s$13.83 213 64.93 19 8.92% 3 15.79% $4.610 T1b/control_10s#2$13.64 193 70.67 25 12.95% 6 24.00% $2.273
这里的点击率是本月最好的,而它们什么也没买到: T1b/utility_6s#2 点击率 16.88%,单次安装仍 要 $1.301,因为千次展示成本是 84.58。
2026-08-16T16:05:25.533743
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
CTR %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CTR by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时 CTR
2026-08-16T16:05:25.630395
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时点击→安装
2026-08-16T16:05:25.679537
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Cost per install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时单次安装成本
2026-08-16T16:05:25.584847
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
USD per 1,000 impressions
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CPM by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时 CPM
B. 分素材 — 三种收入口径
每一列都是自有口径。Meta 只能表达 入账,因此它跟在那一列旁边的括号里,永远不单独成行。付费用户是 去重的人,不是交易笔数。同期群视野 H=24h,本周期覆盖率 100% 。报表时区为 UTC 。
素材 入账 bROAS (Meta) 当日回收 wROAS 同期群@H cROAS 付费用户 付费率 ARPU ARPPU 最大付费用户占比 T1b/utility_6s#2$66.93 0.90 (0.90) $66.93 0.90 $66.93 0.90 3 5.2% 1.154 22.31 52.3% T1b/utility_15s$20.94 0.28 (0.28) $20.94 0.28 $20.94 0.28 3 7.9% 0.551 6.98 52.4% T1b/utility_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — T1b/control_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T1b/control_15s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — T1b/control_10s#2$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — —
16 付费用户门槛(#37):上表中付费用户数不足 16 的行只能看方向,不能看幅度。
整个广告系列六个付费用户,两个 utility_ 格各三个,其他地方一个都没有。它们的 0.90 与 0.28 各自靠 一个人撑着,那个人分别占该格收入的 52.3% 与 52.4%,所以这两格之间的次序就是一个人。该样本量下读不 出来。
2026-08-16T16:05:25.951947
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0
1
2
3
4
5
6
7
payers / installs %
Payer rate by day — every asset (UTC)
T1b/utility_6s#2
T1b/utility_15s
分素材的每日付费率
2026-08-16T16:05:26.002909
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
within-day USD per install
ARPU by day — every asset (UTC)
T1b/utility_6s#2
T1b/utility_15s
分素材的每日 ARPU
2026-08-16T16:05:26.054512
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0
5
10
15
20
within-day USD per payer
ARPPU by day — every asset (UTC)
T1b/utility_6s#2
T1b/utility_15s
分素材的每日 ARPPU
2026-08-16T16:05:26.096480
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.4
0.6
0.8
1.0
revenue / spend
Within-day ROAS by day — every asset (UTC)
T1b/utility_6s#2
T1b/utility_15s
分素材的每日当日回收 ROAS
2026-08-16T16:05:26.162865
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
CTR
click→install
CPI
payer rate *
ARPU *
wROAS *
0.0
0.2
0.4
0.6
0.8
1.0
normalised across the period (1.0 = best of the pack)
* axis sits under the 16-payer bar on most days: direction only, never a level. Assets under 20 installs on the period are not plotted.
Composite — every asset across every measure, 2026-07-23..2026-07-31
T1b/utility_6s#2
T1b/utility_15s
综合图:各素材在各口径上按本周期归一化后的表现
C. 分素材 — 国家结构
⚠ 本表按账户日切,而本页其余各表都在 UTC 时钟上。 Meta 不提供小时 × 国家的交叉表,因此分国家的投放数据只存 在于广告账户自己的 UTC-7 日上,无法重新拼接。它与上面的 A、B 两表相差七小时;本表只读结构,不与它 们比水平。
B 表里某一行读不读得出来,看这里。最后一列是该素材自身 当日回收中来自美国的占比:当它远高于 该素材的美国花费占比时,这条素材的回报是它买到的那块地域的属性。
素材 花费 印度占比 美国占比 其他占比 美国占其自身收入 T1b/utility_6s#2$74.18 0.0% 48.4% 51.6% 100.0% T1b/utility_15s$73.98 0.0% 42.5% 57.5% 52.4% T1b/utility_10s$14.31 0.0% 46.3% 53.7% — T1b/control_6s$13.99 0.0% 39.5% 60.5% — T1b/control_15s$13.83 0.0% 37.4% 62.6% — T1b/control_10s#2$13.64 0.0% 38.8% 61.2% —
T1b/utility_6s#2 把 48.4% 的钱花在美国,当日回收的 100.0% 也来自美国。三个付费用户落在一个国 家里,另外四个国家一个都没有,这点样本读不出一个关于美国市场的结论。
2026-08-16T16:05:25.773655
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
T1b/utility_6s#2
T1b/utility_15s
0
20
40
60
80
100
% of that asset's spend
Where each asset bought — share of its own spend by region, 2026-07-23..2026-07-31 (META days)
US
GB
AU
CA
NZ
VI
other
各素材自身花费的地域结构
2026-08-16T16:05:25.824352
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
T1b/utility_6s#2
T1b/utility_15s
0.0
0.5
1.0
1.5
2.0
2.5
3.0
USD per install
Cost per install by region — every asset, 2026-07-23..2026-07-31 (META days)
US
GB
AU
CA
NZ
各素材分地域的单次安装成本
2026-08-16T16:05:25.871881
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
T1b/utility_6s#2
T1b/utility_15s
0
10
20
30
40
50
60
70
USD
Within-day revenue by region — every asset, 2026-07-23..2026-07-31 (META days)
US
CA
各素材分地域的当日回收 德语区(DE/AT/CH) —— 这一行以下的每一节都是 德语区(DE/AT/CH) 的:当日、分小时、分国家、分广告系列、分素材,都记在 德语区(DE/AT/CH) 自己那本账、自己那些账户上。用上面的按钮切换。
德语区广告系列,2026-07-30 发布并开投,DE+AT 与 CH 各一个广告组,各每日 $75。在本月它只跑了九天里 的两天,花费 $99.98:2,140 次展示、CPM 46.72,23 次安装、单次 $4.347,当日回收 $127.00、回报 1.27,背后两个付费用户。它是本周期里唯一一个回报在 1.00 以上、同时又读不出来的广告系列。
分小时、分国家与汇总
2026-08-16T16:05:28.033077
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
3
6
9
12
15
18
21
hour (UTC)
0
5
10
15
20
25
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Spend by hour
0
3
6
9
12
15
18
21
hour (UTC)
0
100
200
300
400
500
impressions
Impressions and installs by hour
impressions
installs
0
2
4
6
8
installs (Meta Leads)
The period's motion — whole account, 2026-07-23..2026-07-31 (UTC)
本周期走势:分小时花费,以及展示与安装对照
2026-08-16T16:05:28.188350
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
4
8
12
16
20
hour (UTC)
2.90
2.95
3.00
3.05
3.10
3.15
3.20
CTR %
CTR
0
4
8
12
16
20
hour (UTC)
36
37
38
39
USD per 1,000 impressions
CPM
0
4
8
12
16
20
hour (UTC)
25.5
26.0
26.5
27.0
27.5
28.0
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install
0
4
8
12
16
20
hour (UTC)
4.4
4.5
4.6
4.7
4.8
USD
Cost per install
Delivery by hour — German (DE/AT/CH), 2026-07-23..2026-07-31 (UTC)
German (DE/AT/CH) 的分小时 CTR、CPM、点击→安装与 CPI
2026-08-16T16:05:28.306768
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
DE
CH
AT
0
20
40
60
80
USD
spend and installs are a META-day cut (no hour x country grid exists); revenue is the report period
Spend against within-day revenue by country — 2026-07-23..2026-07-31
spend
within-day revenue
分国家的花费与当日回收对照
2026-08-16T16:05:28.345711
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
DE
CH
AT
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
20.0
USD per install
Cost per install by country — 2026-07-23..2026-07-31 (META days)
分国家的单次安装成本
2026-08-16T16:05:28.404446
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
booked
within-day
cohort @ 24h
0
20
40
60
80
100
120
140
USD
1.49
1.27
1.45
the number above each bar is that measure's ROAS against the same period's spend
The three revenue measures, 2026-07-23..2026-07-31 (UTC)
2026-07-23..2026-07-31
两个周期的三个收入口径,柱顶标注 ROAS
2026-08-16T16:05:28.451361
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
2026-07-23..2026-07-31
0
20
40
60
80
100
120
140
USD
the bottom segment IS the within-day figure: the same number seen twice
Booked revenue by how long its payer had been installed (UTC)
paid the day they installed
one day after
two days after
three days after
four or more days after
入账收入按付费用户已安装的天数拆分
A. 分素材 — 投放
分子分母都取自 Meta 的计量口径。我们自己的注册数不进入这张表。
素材 花费 展示 CPM 点击 CTR 安装 点击→安装 CPI DE/de_german_mature_6s$27.40 456 60.09 29 6.36% 8 27.59% $3.425 DE/ch_arab_gulf_6s$21.75 498 43.67 12 2.41% 1 8.33% $21.750 DE/de_arab_gulf_10s$5.79 83 69.76 5 6.02% 1 20.00% $5.790 DE/ch_german_10s$5.20 102 50.98 2 1.96% 0 0.00% — DE/ch_latina_true_6s$5.17 106 48.77 1 0.94% 0 0.00% — DE/de_turkish_mature_10s$4.71 126 37.38 5 3.97% 3 60.00% $1.570 DE/german_6s$4.07 100 40.70 6 6.00% 0 0.00% — DE/ch_german_mature_6s$4.02 103 39.03 0 0.00% 0 — — DE/turkish_mature$3.95 57 69.30 4 7.02% 4 100.00% $0.987 DE/ch_arab_gulf_10s$3.66 146 25.07 1 0.68% 1 100.00% $3.660 DE/german_mature$3.01 89 33.82 6 6.74% 2 33.33% $1.505 DE/de_arab_gulf_6s$2.51 26 96.54 2 7.69% 1 50.00% $2.510 DE/german$1.90 60 31.67 0 0.00% 0 — — DE/latina_true$1.36 21 64.76 1 4.76% 0 0.00% — DE/de_turkish_10s$1.16 30 38.67 6 20.00% 2 33.33% $0.580 DE/de_pool5_15s$1.04 21 49.52 0 0.00% 0 — — DE/de_asian_mature_6s$0.99 37 26.76 0 0.00% 0 — — DE/ch_german_mature_10s$0.76 30 25.33 1 3.33% 0 0.00% — DE/turkish_6s$0.54 24 22.50 1 4.17% 0 0.00% — DE/ch_latina_mature_10s$0.42 7 60.00 0 0.00% 0 — — DE/de_latina_true_6s$0.25 5 50.00 0 0.00% 0 — — DE/ch_german_6s$0.14 4 35.00 0 0.00% 0 — — DE/ch_asian_mature_6s$0.11 3 36.67 1 33.33% 0 0.00% — DE/ch_pool5_15s$0.07 4 17.50 0 0.00% 0 — —
二十四条素材分掉了 $99.98,其中十九条拿到不足 $5。 只有最大的两条点击率读得出来,6.36% 对 2.41%。
2026-08-16T16:05:28.516384
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
CTR %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CTR by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时 CTR
2026-08-16T16:05:28.610175
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时点击→安装
2026-08-16T16:05:28.655760
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Cost per install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时单次安装成本
2026-08-16T16:05:28.565166
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
USD per 1,000 impressions
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CPM by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时 CPM
B. 分素材 — 三种收入口径
每一列都是自有口径。Meta 只能表达 入账,因此它跟在那一列旁边的括号里,永远不单独成行。付费用户是 去重的人,不是交易笔数。同期群视野 H=24h,本周期覆盖率 100% 。报表时区为 UTC 。
素材 入账 bROAS (Meta) 当日回收 wROAS 同期群@H cROAS 付费用户 付费率 ARPU ARPPU 最大付费用户占比 DE/de_german_mature_6s$101.66 3.71 (3.36) $92.36 3.37 $98.10 3.58 1 12.5% 11.545 92.36 100.0% DE/ch_arab_gulf_6s$47.02 2.16 (2.15) $34.65 1.59 $47.02 2.16 1 50.0% 17.323 34.65 100.0% DE/de_arab_gulf_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — DE/ch_german_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/ch_latina_true_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/de_turkish_mature_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — DE/german_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/ch_german_mature_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/turkish_mature$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — DE/ch_arab_gulf_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — DE/german_mature$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — DE/de_arab_gulf_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — DE/german$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/latina_true$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/de_turkish_10s$0.00 0.00 (4.30) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — — DE/de_pool5_15s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/de_asian_mature_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/ch_german_mature_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/turkish_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/ch_latina_mature_10s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/de_latina_true_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/ch_german_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/ch_asian_mature_6s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — DE/ch_pool5_15s$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — —
16 付费用户门槛(#37):上表中付费用户数不足 16 的行只能看方向,不能看幅度。
整个广告系列两个付费用户,分别落在两条素材上,各占该素材收入的 100.0%。3.37 与 1.59 就是两个人,第 5 节里这个广告系列的 1.27 是同样这两个人高了一层。该样本量下读不出来。7 月这次德语区读数真正立住的 东西写在 DECISION_LOG.md #44 里,靠的是投放数据:德语区的展示价格是印度的几倍,而这块地域每次安 装赚到的钱足以撑住它,只是素材包里没有一条到得了盈亏平衡。
2026-08-16T16:05:28.865775
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
payers / installs %
Payer rate by day — every asset (UTC)
分素材的每日付费率
2026-08-16T16:05:28.903336
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
within-day USD per install
ARPU by day — every asset (UTC)
分素材的每日 ARPU
2026-08-16T16:05:28.942014
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
within-day USD per payer
ARPPU by day — every asset (UTC)
分素材的每日 ARPPU
2026-08-16T16:05:28.981124
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
revenue / spend
Within-day ROAS by day — every asset (UTC)
分素材的每日当日回收 ROAS
2026-08-16T16:05:29.043539
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
CTR
click→install
CPI
payer rate *
ARPU *
wROAS *
0.0
0.2
0.4
0.6
0.8
1.0
normalised across the period (1.0 = best of the pack)
* axis sits under the 16-payer bar on most days: direction only, never a level. Assets under 20 installs on the period are not plotted.
Composite — every asset across every measure, 2026-07-23..2026-07-31
综合图:各素材在各口径上按本周期归一化后的表现
C. 分素材 — 国家结构
⚠ 本表按账户日切,而本页其余各表都在 UTC 时钟上。 Meta 不提供小时 × 国家的交叉表,因此分国家的投放数据只存 在于广告账户自己的 UTC-7 日上,无法重新拼接。它与上面的 A、B 两表相差七小时;本表只读结构,不与它 们比水平。
B 表里某一行读不读得出来,看这里。最后一列是该素材自身 当日回收中来自美国的占比:当它远高于 该素材的美国花费占比时,这条素材的回报是它买到的那块地域的属性。
素材 花费 印度占比 美国占比 其他占比 美国占其自身收入 DE/de_german_mature_6s$27.40 0.0% 0.0% 100.0% 0.0% DE/ch_arab_gulf_6s$21.75 0.0% 0.0% 100.0% 0.0% DE/de_arab_gulf_10s$5.79 0.0% 0.0% 100.0% — DE/ch_german_10s$5.20 0.0% 0.0% 100.0% — DE/ch_latina_true_6s$5.17 0.0% 0.0% 100.0% — DE/de_turkish_mature_10s$4.71 0.0% 0.0% 100.0% — DE/german_6s$4.07 0.0% 0.0% 100.0% — DE/ch_german_mature_6s$4.02 0.0% 0.0% 100.0% — DE/turkish_mature$3.95 0.0% 0.0% 100.0% — DE/ch_arab_gulf_10s$3.66 0.0% 0.0% 100.0% — DE/german_mature$3.01 0.0% 0.0% 100.0% — DE/de_arab_gulf_6s$2.51 0.0% 0.0% 100.0% — DE/german$1.90 0.0% 0.0% 100.0% — DE/latina_true$1.36 0.0% 0.0% 100.0% — DE/de_turkish_10s$1.16 0.0% 0.0% 100.0% — DE/de_pool5_15s$1.04 0.0% 0.0% 100.0% — DE/de_asian_mature_6s$0.99 0.0% 0.0% 100.0% — DE/ch_german_mature_10s$0.76 0.0% 0.0% 100.0% — DE/turkish_6s$0.54 0.0% 0.0% 100.0% — DE/ch_latina_mature_10s$0.42 0.0% 0.0% 100.0% — DE/de_latina_true_6s$0.25 0.0% 0.0% 100.0% — DE/ch_german_6s$0.14 0.0% 0.0% 100.0% — DE/ch_asian_mature_6s$0.11 0.0% 0.0% 100.0% — DE/ch_pool5_15s$0.07 0.0% 0.0% 100.0% —
每一行都是 100.0% 其他,这是这个广告系列的定向要求的。本表分不出来的部分在第 3 节:德国 $60.82、回 报 1.67,瑞士 $43.05、回报 0.80,各自背后一个付费用户。
2026-08-16T16:05:28.738532
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
20
40
60
80
100
% of that asset's spend
Where each asset bought — share of its own spend by region, 2026-07-23..2026-07-31 (META days)
DE
CH
AT
other
各素材自身花费的地域结构
2026-08-16T16:05:28.766757
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0.0
0.2
0.4
0.6
0.8
1.0
USD per install
Cost per install by region — every asset, 2026-07-23..2026-07-31 (META days)
各素材分地域的单次安装成本
2026-08-16T16:05:28.791976
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0.0
0.2
0.4
0.6
0.8
1.0
USD
Within-day revenue by region — every asset, 2026-07-23..2026-07-31 (META days)
各素材分地域的当日回收 T1(7 月 23 日) —— 这一行以下的每一节都是 T1(7 月 23 日) 的:当日、分小时、分国家、分广告系列、分素材,都记在 T1(7 月 23 日) 自己那本账、自己那些账户上。用上面的按钮切换。
第一轮的英语一档,来自 2026-07-23 那一轮。它花了 $72.39,买了 1,296 次展示、CPM 55.86,产出四次安 装,单次 $18.098。它没有付费用户,三个收入口径都是 $0.00。它凭花费占比拿到了一个标签页,此外什么也 没拿到:这是一个买到了展示、之后几乎什么都没买到的广告系列。
分小时、分国家与汇总
2026-08-16T16:05:30.903806
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
3
6
9
12
15
18
21
hour (UTC)
0
5
10
15
20
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Spend by hour
0
3
6
9
12
15
18
21
hour (UTC)
0
50
100
150
200
250
300
impressions
Impressions and installs by hour
impressions
installs
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
installs (Meta Leads)
The period's motion — whole account, 2026-07-23..2026-07-31 (UTC)
本周期走势:分小时花费,以及展示与安装对照
2026-08-16T16:05:31.051686
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
4
8
12
16
20
hour (UTC)
−0.04
−0.02
0.00
0.02
0.04
CTR %
CTR
0
4
8
12
16
20
hour (UTC)
−0.04
−0.02
0.00
0.02
0.04
USD per 1,000 impressions
CPM
0
4
8
12
16
20
hour (UTC)
−0.04
−0.02
0.00
0.02
0.04
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install
0
4
8
12
16
20
hour (UTC)
−0.04
−0.02
0.00
0.02
0.04
USD
Cost per install
Delivery by hour — T1 (Jul 23), 2026-07-23..2026-07-31 (UTC)
T1 (Jul 23) 的分小时 CTR、CPM、点击→安装与 CPI
2026-08-16T16:05:31.171172
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
US
GB
CA
AU
NZ
VI
0
5
10
15
20
25
30
35
USD
spend and installs are a META-day cut (no hour x country grid exists); revenue is the report period
Spend against within-day revenue by country — 2026-07-23..2026-07-31
spend
within-day revenue
分国家的花费与当日回收对照
2026-08-16T16:05:31.208378
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
GB
CA
NZ
0
5
10
15
20
USD per install
Cost per install by country — 2026-07-23..2026-07-31 (META days)
分国家的单次安装成本
2026-08-16T16:05:31.261730
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
booked
within-day
cohort @ 24h
0
2
4
6
8
USD
0.13
0.00
0.13
the number above each bar is that measure's ROAS against the same period's spend
The three revenue measures, 2026-07-23..2026-07-31 (UTC)
2026-07-23..2026-07-31
两个周期的三个收入口径,柱顶标注 ROAS
2026-08-16T16:05:31.302518
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
2026-07-23..2026-07-31
0
2
4
6
8
USD
the bottom segment IS the within-day figure: the same number seen twice
Booked revenue by how long its payer had been installed (UTC)
paid the day they installed
one day after
two days after
three days after
four or more days after
入账收入按付费用户已安装的天数拆分
A. 分素材 — 投放
分子分母都取自 Meta 的计量口径。我们自己的注册数不进入这张表。
素材 花费 展示 CPM 点击 CTR 安装 点击→安装 CPI T1a/utility_sexy$30.97 559 55.40 24 4.29% 0 0.00% — T1a/control_sexy$28.95 349 82.95 40 11.46% 4 10.00% $7.237 T1a/control_sexy#4$6.56 178 36.85 4 2.25% 0 0.00% — T1a/utility_sexy#4$5.91 210 28.14 3 1.43% 0 0.00% —
四条素材里有三条一次安装也没产出。第四条 T1a/control_sexy 产出四次,单次 $7.237,背后是 349 次以 82.95 千次成本买来的展示。本该被拿来比较的两条实验臂 utility_sexy 与 control_sexy 花了 $30.97 与 $28.95,产出零次对四次安装,这是四个事件的差别。
2026-08-16T16:05:31.364444
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
CTR %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CTR by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时 CTR
2026-08-16T16:05:31.457497
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时点击→安装
2026-08-16T16:05:31.504632
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Cost per install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时单次安装成本
2026-08-16T16:05:31.409699
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
USD per 1,000 impressions
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CPM by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时 CPM
B. 分素材 — 三种收入口径
每一列都是自有口径。Meta 只能表达 入账,因此它跟在那一列旁边的括号里,永远不单独成行。付费用户是 去重的人,不是交易笔数。同期群视野 H=24h,本周期覆盖率 100% 。报表时区为 UTC 。
素材 入账 bROAS (Meta) 当日回收 wROAS 同期群@H cROAS 付费用户 付费率 ARPU ARPPU 最大付费用户占比 T1a/utility_sexy$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — T1a/control_sexy$9.39 0.32 (0.32) $0.00 0.00 $9.39 0.32 0 0.0% 0.000 — — T1a/control_sexy#4$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — — T1a/utility_sexy#4$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 — — — —
16 付费用户门槛(#37):上表中付费用户数不足 16 的行只能看方向,不能看幅度。
$9.39 的入账收入,对上 $72.39 的花费,其中没有一分是当日回收的,整个广告系列没有付费用户。平时那句 提醒在这里用不上,因为这里没有任何数字需要提醒。
2026-08-16T16:05:31.720812
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
payers / installs %
Payer rate by day — every asset (UTC)
分素材的每日付费率
2026-08-16T16:05:31.758779
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
within-day USD per install
ARPU by day — every asset (UTC)
分素材的每日 ARPU
2026-08-16T16:05:31.797515
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
within-day USD per payer
ARPPU by day — every asset (UTC)
分素材的每日 ARPPU
2026-08-16T16:05:31.844937
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
revenue / spend
Within-day ROAS by day — every asset (UTC)
分素材的每日当日回收 ROAS
2026-08-16T16:05:31.907908
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
CTR
click→install
CPI
payer rate *
ARPU *
wROAS *
0.0
0.2
0.4
0.6
0.8
1.0
normalised across the period (1.0 = best of the pack)
* axis sits under the 16-payer bar on most days: direction only, never a level. Assets under 20 installs on the period are not plotted.
Composite — every asset across every measure, 2026-07-23..2026-07-31
综合图:各素材在各口径上按本周期归一化后的表现
C. 分素材 — 国家结构
⚠ 本表按账户日切,而本页其余各表都在 UTC 时钟上。 Meta 不提供小时 × 国家的交叉表,因此分国家的投放数据只存 在于广告账户自己的 UTC-7 日上,无法重新拼接。它与上面的 A、B 两表相差七小时;本表只读结构,不与它 们比水平。
B 表里某一行读不读得出来,看这里。最后一列是该素材自身 当日回收中来自美国的占比:当它远高于 该素材的美国花费占比时,这条素材的回报是它买到的那块地域的属性。
素材 花费 印度占比 美国占比 其他占比 美国占其自身收入 T1a/utility_sexy$30.97 0.0% 51.2% 48.8% — T1a/control_sexy$28.95 0.0% 57.5% 42.5% — T1a/control_sexy#4$6.56 0.0% 21.2% 78.8% — T1a/utility_sexy#4$5.91 0.0% 32.0% 68.0% —
每条素材有五分之一到五分之三的钱花在美国,而最后一列每一行都是空的,因为根本没有收入可以归。
2026-08-16T16:05:31.594153
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
20
40
60
80
100
% of that asset's spend
Where each asset bought — share of its own spend by region, 2026-07-23..2026-07-31 (META days)
US
GB
CA
AU
NZ
VI
other
各素材自身花费的地域结构
2026-08-16T16:05:31.623785
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0.0
0.2
0.4
0.6
0.8
1.0
USD per install
Cost per install by region — every asset, 2026-07-23..2026-07-31 (META days)
各素材分地域的单次安装成本
2026-08-16T16:05:31.650038
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0.0
0.2
0.4
0.6
0.8
1.0
USD
Within-day revenue by region — every asset, 2026-07-23..2026-07-31 (META days)
各素材分地域的当日回收 T2(7 月 23 日) —— 这一行以下的每一节都是 T2(7 月 23 日) 的:当日、分小时、分国家、分广告系列、分素材,都记在 T2(7 月 23 日) 自己那本账、自己那些账户上。用上面的按钮切换。
第一轮的中间一档。$55.25 买了 2,680 次展示、CPM 20.62,35 次安装、单次 $1.579,当日回收 $9.11 、回报 0.16,背后一个付费用户。
分小时、分国家与汇总
2026-08-16T16:05:33.743508
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
3
6
9
12
15
18
21
hour (UTC)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Spend by hour
0
3
6
9
12
15
18
21
hour (UTC)
80
100
120
140
160
180
impressions
Impressions and installs by hour
impressions
installs
0
1
2
3
4
5
installs (Meta Leads)
The period's motion — whole account, 2026-07-23..2026-07-31 (UTC)
本周期走势:分小时花费,以及展示与安装对照
2026-08-16T16:05:33.890951
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
4
8
12
16
20
hour (UTC)
−0.04
−0.02
0.00
0.02
0.04
CTR %
CTR
0
4
8
12
16
20
hour (UTC)
−0.04
−0.02
0.00
0.02
0.04
USD per 1,000 impressions
CPM
0
4
8
12
16
20
hour (UTC)
−0.04
−0.02
0.00
0.02
0.04
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install
0
4
8
12
16
20
hour (UTC)
−0.04
−0.02
0.00
0.02
0.04
USD
Cost per install
Delivery by hour — T2 (Jul 23), 2026-07-23..2026-07-31 (UTC)
T2 (Jul 23) 的分小时 CTR、CPM、点击→安装与 CPI
2026-08-16T16:05:34.010271
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
MY
MX
AE
SA
ZA
0.0
2.5
5.0
7.5
10.0
12.5
15.0
17.5
USD
spend and installs are a META-day cut (no hour x country grid exists); revenue is the report period
Spend against within-day revenue by country — 2026-07-23..2026-07-31
spend
within-day revenue
分国家的花费与当日回收对照
2026-08-16T16:05:34.053362
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
MY
MX
AE
SA
ZA
0
1
2
3
4
5
6
7
8
USD per install
Cost per install by country — 2026-07-23..2026-07-31 (META days)
分国家的单次安装成本
2026-08-16T16:05:34.111758
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
booked
within-day
cohort @ 24h
0
2
4
6
8
USD
0.16
0.16
0.16
the number above each bar is that measure's ROAS against the same period's spend
The three revenue measures, 2026-07-23..2026-07-31 (UTC)
2026-07-23..2026-07-31
两个周期的三个收入口径,柱顶标注 ROAS
2026-08-16T16:05:34.151417
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
2026-07-23..2026-07-31
0
2
4
6
8
USD
the bottom segment IS the within-day figure: the same number seen twice
Booked revenue by how long its payer had been installed (UTC)
paid the day they installed
one day after
two days after
three days after
four or more days after
入账收入按付费用户已安装的天数拆分
A. 分素材 — 投放
分子分母都取自 Meta 的计量口径。我们自己的注册数不进入这张表。
素材 花费 展示 CPM 点击 CTR 安装 点击→安装 CPI T2a/control_sexy#2$28.18 1,303 21.63 121 9.29% 13 10.74% $2.168 T2a/utility_sexy#3$27.07 1,377 19.66 175 12.71% 22 12.57% $1.230
这是本月唯一一处两条实验臂花钱相当、并且两条都有投放的地方。 utility 的点击率 12.71%, control 9.29%,后面站着的一共三十五次安装。
2026-08-16T16:05:34.213453
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
CTR %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CTR by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时 CTR
2026-08-16T16:05:34.307435
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时点击→安装
2026-08-16T16:05:34.359676
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Cost per install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时单次安装成本
2026-08-16T16:05:34.261179
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
USD per 1,000 impressions
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CPM by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时 CPM
B. 分素材 — 三种收入口径
每一列都是自有口径。Meta 只能表达 入账,因此它跟在那一列旁边的括号里,永远不单独成行。付费用户是 去重的人,不是交易笔数。同期群视野 H=24h,本周期覆盖率 100% 。报表时区为 UTC 。
素材 入账 bROAS (Meta) 当日回收 wROAS 同期群@H cROAS 付费用户 付费率 ARPU ARPPU 最大付费用户占比 T2a/control_sexy#2$9.11 0.32 (0.32) $9.11 0.32 $9.11 0.32 1 7.7% 0.701 9.11 100.0% T2a/utility_sexy#3$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — —
16 付费用户门槛(#37):上表中付费用户数不足 16 的行只能看方向,不能看幅度。
整个广告系列一个付费用户,而它落在投放较差的那一条上。该样本量下读不出来,两个方向都读不出来:这么 大的一个投放优势,和一个人带来的收入劣势,是两个不同的问题,而只有第一个有样本。
2026-08-16T16:05:34.605644
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.00
0.01
0.02
0.03
0.04
0.05
payers / installs %
Payer rate by day — every asset (UTC)
T2a/utility_sexy#3
分素材的每日付费率
2026-08-16T16:05:34.647788
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.00
0.01
0.02
0.03
0.04
0.05
within-day USD per install
ARPU by day — every asset (UTC)
T2a/utility_sexy#3
分素材的每日 ARPU
2026-08-16T16:05:34.689960
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
within-day USD per payer
ARPPU by day — every asset (UTC)
分素材的每日 ARPPU
2026-08-16T16:05:34.728990
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
−0.04
−0.02
0.00
0.02
0.04
revenue / spend
Within-day ROAS by day — every asset (UTC)
T2a/utility_sexy#3
分素材的每日当日回收 ROAS
2026-08-16T16:05:34.792308
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
CTR
click→install
CPI
payer rate *
ARPU *
wROAS *
0.0
0.2
0.4
0.6
0.8
1.0
normalised across the period (1.0 = best of the pack)
* axis sits under the 16-payer bar on most days: direction only, never a level. Assets under 20 installs on the period are not plotted.
Composite — every asset across every measure, 2026-07-23..2026-07-31
T2a/utility_sexy#3
综合图:各素材在各口径上按本周期归一化后的表现
C. 分素材 — 国家结构
⚠ 本表按账户日切,而本页其余各表都在 UTC 时钟上。 Meta 不提供小时 × 国家的交叉表,因此分国家的投放数据只存 在于广告账户自己的 UTC-7 日上,无法重新拼接。它与上面的 A、B 两表相差七小时;本表只读结构,不与它 们比水平。
B 表里某一行读不读得出来,看这里。最后一列是该素材自身 当日回收中来自美国的占比:当它远高于 该素材的美国花费占比时,这条素材的回报是它买到的那块地域的属性。
素材 花费 印度占比 美国占比 其他占比 美国占其自身收入 T2a/control_sexy#2$28.18 0.0% 0.0% 100.0% 0.0% T2a/utility_sexy#3$27.07 0.0% 0.0% 100.0% —
两行都是 100.0% 其他,没有印度也没有美国,所以两条实验臂买的是同一块地域,上面那处投放差异不是结构 差异。
2026-08-16T16:05:34.448643
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
T2a/utility_sexy#3
0
20
40
60
80
100
% of that asset's spend
Where each asset bought — share of its own spend by region, 2026-07-23..2026-07-31 (META days)
MY
MX
AE
SA
ZA
other
各素材自身花费的地域结构
2026-08-16T16:05:34.489802
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
T2a/utility_sexy#3
0.0
0.5
1.0
1.5
2.0
2.5
3.0
USD per install
Cost per install by region — every asset, 2026-07-23..2026-07-31 (META days)
MY
MX
AE
SA
ZA
各素材分地域的单次安装成本
2026-08-16T16:05:34.523460
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
T2a/utility_sexy#3
0.0
0.2
0.4
0.6
0.8
1.0
USD
Within-day revenue by region — every asset, 2026-07-23..2026-07-31 (META days)
各素材分地域的当日回收 T3(7 月 23 日) —— 这一行以下的每一节都是 T3(7 月 23 日) 的:当日、分小时、分国家、分广告系列、分素材,都记在 T3(7 月 23 日) 自己那本账、自己那些账户上。用上面的按钮切换。
第一轮的走量一档,买印度与东南亚。$40.68 买了 5,902 次展示、CPM 6.89,是本月第二便宜的;118 次安装 、单次 $0.345,当日回收 $67.82、回报 1.67,背后五个付费用户。
分小时、分国家与汇总
2026-08-16T16:05:36.665921
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
3
6
9
12
15
18
21
hour (UTC)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Spend by hour
0
3
6
9
12
15
18
21
hour (UTC)
100
200
300
400
500
600
impressions
Impressions and installs by hour
impressions
installs
0
2
4
6
8
10
installs (Meta Leads)
The period's motion — whole account, 2026-07-23..2026-07-31 (UTC)
本周期走势:分小时花费,以及展示与安装对照
2026-08-16T16:05:36.824455
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
4
8
12
16
20
hour (UTC)
4.90
4.95
5.00
5.05
5.10
5.15
CTR %
CTR
0
4
8
12
16
20
hour (UTC)
3.5
3.6
3.7
3.8
3.9
4.0
4.1
USD per 1,000 impressions
CPM
0
4
8
12
16
20
hour (UTC)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install
0
4
8
12
16
20
hour (UTC)
2.40
2.45
2.50
2.55
2.60
USD
Cost per install
Delivery by hour — T3 (Jul 23), 2026-07-23..2026-07-31 (UTC)
T3 (Jul 23) 的分小时 CTR、CPM、点击→安装与 CPI
2026-08-16T16:05:36.947022
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
IN
ID
PH
PK
NG
0
10
20
30
40
50
60
USD
spend and installs are a META-day cut (no hour x country grid exists); revenue is the report period
Spend against within-day revenue by country — 2026-07-23..2026-07-31
spend
within-day revenue
分国家的花费与当日回收对照
2026-08-16T16:05:36.984542
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
IN
ID
PH
PK
0.0
0.1
0.2
0.3
0.4
0.5
USD per install
Cost per install by country — 2026-07-23..2026-07-31 (META days)
分国家的单次安装成本
2026-08-16T16:05:37.038836
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
booked
within-day
cohort @ 24h
0
20
40
60
80
USD
2.30
1.67
1.75
the number above each bar is that measure's ROAS against the same period's spend
The three revenue measures, 2026-07-23..2026-07-31 (UTC)
2026-07-23..2026-07-31
两个周期的三个收入口径,柱顶标注 ROAS
2026-08-16T16:05:37.083271
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
2026-07-23..2026-07-31
0
20
40
60
80
USD
the bottom segment IS the within-day figure: the same number seen twice
Booked revenue by how long its payer had been installed (UTC)
paid the day they installed
one day after
two days after
three days after
four or more days after
入账收入按付费用户已安装的天数拆分
A. 分素材 — 投放
分子分母都取自 Meta 的计量口径。我们自己的注册数不进入这张表。
素材 花费 展示 CPM 点击 CTR 安装 点击→安装 CPI T3a/utility_sexy#2$28.90 4,173 6.93 478 11.45% 114 23.85% $0.254 T3a/control_sexy#3$11.78 1,729 6.81 114 6.59% 4 3.51% $2.945
与 07-23 那一轮的中间一档形状相同、方向相同,只是量更大。在千次成本基本一样的展示上,6.93 对 6.81 ,utility 的点击率是 11.45%,control 是 6.59%;点击转安装 23.85% 对 3.51%,最后落到单次安装 $0.254 对 $2.945。这里花费没有对齐,$28.90 对 $11.78,所以读比率,不要读合计。
2026-08-16T16:05:37.143613
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
CTR %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CTR by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时 CTR
2026-08-16T16:05:37.234196
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
installs / clicks %
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Click → install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时点击→安装
2026-08-16T16:05:37.280557
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
USD
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
Cost per install by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时单次安装成本
2026-08-16T16:05:37.189417
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
0
2
4
6
8
10
12
14
16
18
20
22
hour (UTC)
0.0
0.2
0.4
0.6
0.8
1.0
USD per 1,000 impressions
hour of day, pooled across the 9 days of the period — a pooling, not a timeline
CPM by hour — every asset, 2026-07-23..2026-07-31 (UTC)
分素材的分小时 CPM
B. 分素材 — 三种收入口径
每一列都是自有口径。Meta 只能表达 入账,因此它跟在那一列旁边的括号里,永远不单独成行。付费用户是 去重的人,不是交易笔数。同期群视野 H=24h,本周期覆盖率 100% 。报表时区为 UTC 。
素材 入账 bROAS (Meta) 当日回收 wROAS 同期群@H cROAS 付费用户 付费率 ARPU ARPPU 最大付费用户占比 T3a/utility_sexy#2$93.57 3.24 (2.23) $67.82 2.35 $71.38 2.47 5 3.7% 0.502 13.56 51.7% T3a/control_sexy#3$0.00 0.00 (0.00) $0.00 0.00 $0.00 0.00 0 0.0% 0.000 — —
16 付费用户门槛(#37):上表中付费用户数不足 16 的行只能看方向,不能看幅度。
五个付费用户上的 2.35,其中 51.7% 来自一个人。该样本量下读不出来;第 5 节里这个广告系列的 1.67,就 是这一格被那条什么也没产出的实验臂摊薄之后的结果。
2026-08-16T16:05:37.523350
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.0
0.5
1.0
1.5
2.0
2.5
payers / installs %
Payer rate by day — every asset (UTC)
T3a/utility_sexy#2
分素材的每日付费率
2026-08-16T16:05:37.572985
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
within-day USD per install
ARPU by day — every asset (UTC)
T3a/utility_sexy#2
分素材的每日 ARPU
2026-08-16T16:05:37.619630
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
0
2
4
6
8
10
12
14
16
within-day USD per payer
ARPPU by day — every asset (UTC)
T3a/utility_sexy#2
分素材的每日 ARPPU
2026-08-16T16:05:37.668676
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
07-23
07-24
07-25
07-26
07-27
07-28
07-29
07-30
07-31
day (UTC)
1.700
1.725
1.750
1.775
1.800
1.825
1.850
1.875
revenue / spend
Within-day ROAS by day — every asset (UTC)
T3a/utility_sexy#2
分素材的每日当日回收 ROAS
2026-08-16T16:05:37.736203
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
CTR
click→install
CPI
payer rate *
ARPU *
wROAS *
0.0
0.2
0.4
0.6
0.8
1.0
normalised across the period (1.0 = best of the pack)
* axis sits under the 16-payer bar on most days: direction only, never a level. Assets under 20 installs on the period are not plotted.
Composite — every asset across every measure, 2026-07-23..2026-07-31
T3a/utility_sexy#2
综合图:各素材在各口径上按本周期归一化后的表现
C. 分素材 — 国家结构
⚠ 本表按账户日切,而本页其余各表都在 UTC 时钟上。 Meta 不提供小时 × 国家的交叉表,因此分国家的投放数据只存 在于广告账户自己的 UTC-7 日上,无法重新拼接。它与上面的 A、B 两表相差七小时;本表只读结构,不与它 们比水平。
B 表里某一行读不读得出来,看这里。最后一列是该素材自身 当日回收中来自美国的占比:当它远高于 该素材的美国花费占比时,这条素材的回报是它买到的那块地域的属性。
素材 花费 印度占比 美国占比 其他占比 美国占其自身收入 T3a/utility_sexy#2$28.90 79.4% 0.0% 20.6% 0.0% T3a/control_sexy#3$11.78 57.8% 0.0% 42.2% —
两条实验臂买到的结构并不一样:印度占比 79.4% 对 57.8%,其余落在 T3 的其他国家。印度是这几个国家里 最便宜、转化也最好的一个,所以上面那处投放差距里,有一部分是竞价把两条实验臂放到了不同的地方。
2026-08-16T16:05:37.370445
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
T3a/utility_sexy#2
0
20
40
60
80
100
% of that asset's spend
Where each asset bought — share of its own spend by region, 2026-07-23..2026-07-31 (META days)
IN
ID
PH
PK
NG
other
各素材自身花费的地域结构
2026-08-16T16:05:37.407664
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
T3a/utility_sexy#2
0.0
0.1
0.2
0.3
0.4
USD per install
Cost per install by region — every asset, 2026-07-23..2026-07-31 (META days)
IN
ID
PH
PK
各素材分地域的单次安装成本
2026-08-16T16:05:37.441970
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
T3a/utility_sexy#2
0
10
20
30
40
50
60
USD
Within-day revenue by region — every asset, 2026-07-23..2026-07-31 (META days)
IN
PH
各素材分地域的当日回收
7 口径与边界
⚠ 口径校验落在区间之外,这里不做四舍五入把它凑进来。 在账户时区上(校验只在这个时钟上有效), 我们自己的记录是 Meta 入账 $3,242.34 的 1.07 倍,有记录的区间是 0.91 到 1.06。在本报告自己的 UTC 时 钟上,同一个比值是 1.08,而它根本不构成一次有效的校验,因为 Meta 的转化金额盖在转化发生的小时上, 那是分小时导出里唯一一列无法跨两个账户日重新拼接的。两个数字都碰不到当日回收口径,后者从不读 Meta 那一列;它们限定的是任何把我们的数字与 Meta 放在一起说的句子。
⚠ 付款关联的修复够不到 7 月。 这个口径会丢掉任何无法关联到人的付款,2026-08-16 那次修复靠的是 按账户 UUID 关联,而注册推送从 2026-08-13 18:52 UTC 才开始带这个字段。7 月这批人的注册远早于此,修 复伸不过去,早期数据流当时归不上的那些付款仍然归不上(asset_performance_report_2026-08-15.md)。 7 月缺掉的部分会把收入读低,而口径校验读出来高于 Meta,所以这两件事不会叠加。
德语区投放里 ch_ 与 de_ 两个前缀标的是投放地域的切分。 这两组用的是同样的镜头、同样的字 幕、同样的配音,只有背景音乐不同(ad_ops/dach_stage2_asset_proposal_2026-07-30.md)。 DE/ch_arab_gulf_6s 买了 498 次展示、一次安装,单次 $21.750。
⚠ 本页没有任何一处是同预算区间的。 预算区间是广告组日预算没有变动的一段时间,而在这个账户上预 算是一个地域旋钮(DECISION_LOG.md #47)。估计器手工记录的预算区间表只覆盖旧广告系列和德语区投 放,六个 T1/T2/T3 广告系列一张也没有。在这九天里,单是旧广告系列就从每日 $150 动到 $250、$500 与 $875,而 07-23 那一轮的广告组从 $5 提到 $20,时间点没有任何记录。因此上面每一个分广 告系列和分素材的数字,都跨越了预算变动。
⚠ 分国家一节与所有 C 表走的都是账户自己的日 ,比本页其余部分晚七小时。在旧广告系列上,这个窗口 伸进了 2026-08-01 UTC,给出 $408.99 与 $150.33,而 A 表给出的是 $312.93 与 $8.64。它们只读结构,不 读水平。
这里的同期群已经走满,本系列里没有哪一份日报能这样说。 5,069 次安装在付款数据截止时间上的年龄是 371.0 到 567.3 小时,H=24 覆盖率 100%,所以 $2,697.93 已经冻结。同期群那一列不会再往上走,也不需要挂 任何覆盖率说明。
这个月测的那套分组,在这个月里就已经停用了。 DECISION_LOG.md #36 在 2026-07-26 放弃了 T1/T2/T3 这套切分:这些分组是按假定成本凑出来的,按实测付费率读出来次序是反的,而且每一组内部把性 质完全不同的国家平均在了一起。这六个广告系列写在这里,是因为它们花了钱。
2026-07-25 那一轮回答不了它自己提的问题,本报告继承了这一点。 十八个格,每格一个广告组,任何一 处都没有重复,16 付费用户门槛就是从这里来的(DECISION_LOG.md #37)。九天的买量下来,八个广告系列 里有三个、整份素材清单里有五条,过得了这道门槛。
本页贯穿着两套安装数,它们从不交叉相除。 Meta 的 4,561 对我们自己的 5,069 次注册。投放漏斗的分子分 母都用 Meta 的;付费率与每一个收入口径的分母用我们自己的。
本页在账户层带 18 张图,八个标签页里各再带 18 张。 每一张在发布时都以 SVG 内联进页面,所以即使 一个标签页被折起来,它的图也仍然随页面一起发出去,页面因此很大。把它们从标签页里搬出去,等于把本格 式已经删掉的图表附录又请回来,所以它们留在原处。
8 要把这些问题定下来,需要什么
印度的 1.00 到底算不算一门生意。 它是本月唯一一格背后有足够多付费用户、可以当水平来读的,而它 正好停在为它付出的钱上,还没有扣掉应用商店的分成。7 月内部没有任何东西能改变这一点;8 月的报告是第 一份能说出它往哪边走的。
7 月有没有哪条素材在收入上与另一条不同。 五条素材过了 16 付费用户门槛,其中四条在同一个广告系 列里,而且没有一条是同预算区间的。这个账户自己的可判定性算法要求每条实验臂上千次安装,远高于一轮九 天、十八个格的测试能产出的量,所以老实的答案是:7 月测的不是素材,是地域。
那一轮分组测试到底花了多少。 第 5 节把六个分组广告系列逐个与旧广告系列、德语区投放放在一起标价 。那张表就是完整的账,读它不需要第二个周期;它说不了的是同样这笔钱如果放进一个不设限的全球买量里会 有什么结果,而账户在 2026-07-30 转向的正是那个形态。
9 交给下一次读数的东西
8 月是本系列里第一个身后有东西的周期,而这份报告就是那个东西。它交出去的是一个水平:账户 CPI $0.538,当日回收回报 0.95,$2,344.02 背后是 233 个付费用户的付费率,以及一个已经定住的同期群数字。 任何 8 月的对比都对着这四个数来做。
在做那次对比时,有三件事要放在心里。这九天并不是同一个账户的九天,因为八个广告系列里有六个在这九天 之内被关掉或改过结构。旧广告系列只跑了其中两天,而且在这两天里把预算提到了原来的几倍,所以拿 7 月 的它去比 8 月的它,是拿两天的爬坡去比另一段爬坡。还有,分国家的切分两头都走账户自己的时区,所以到 了 8 月它仍然只是一个结构读数。
10 复现
报表时区为 UTC 。B1 的导出盖的是它自己账户时区(UTC-7)的时间戳,因此上面每一行投放数据都由两个 Meta 日重新拼接。分国家的表与所有 C 表不做拼接,并在出现的地方标注。
运行目录 _runs/2026-08-16_periods/,由 generate.sh 生成周期 2026-07-23 → 2026-07-31 UTC,九天,之前没有任何周期 配置 _runs/2026-08-16_receipt_series/run_config_full.json;口径校验走 run_config_full_meta.json付款数据截止时间 报告为 2026-08-16 10:59 UTC,口径校验为账户时区的 2026-08-16 03:59 Meta 导出 分小时按账户日取自 run config 的存档,已验证带满 24 个小时桶;分日与国家×日取自 _runs/2026-08-16_2026-08-15_daily_auto/exports/ 注册与付款 _runs/2026-08-16_2026-08-15_daily_auto/lum汇率 _runs/2026-08-02_today_read/fx_rates.json预算区间 没有任何预算区间表覆盖 T1/T2/T3 广告系列;旧广告系列的表是 ad_ops/asset_estimator/cohorts.py 里的 REGIMES_260730
7 月账户花掉、而本报告不统计的部分。 2026-07-15 至 2026-07-21,账户上跑的是 youguqi_ 黑哈测 试,$477.52 ,投的是另一个 App,走的是借来的归因栈。2026-07-30 与 2026-07-31,账户上还跑了 0730_starfall_,$71.62 ,是替第三方的 App 代投。两笔都不是这个 App 的钱,在 run config 里都 标为跳过;账户自己的投放导出里 7 月花费比本页的 $2,455.00 多,就是因为它们。
CFG=_runs/2026-08-16_receipt_series/run_config_full.json
python -m ad_ops.measures --config $CFG --from 2026-07-23 --to 2026-07-31
python -m ad_ops.report_tables --config $CFG --from 2026-07-23 --to 2026-07-31
python -m ad_ops.report_tables --config $CFG --from 2026-07-23 --to 2026-07-31 --campaign-table
python -m ad_ops.report_tables --config $CFG --from 2026-07-23 --to 2026-07-31 --campaign t3_0725
python -m ad_ops.report_charts --config $CFG --from 2026-07-23 --to 2026-07-31 \
--out ad_ops/figures/month_2026-07 --md _runs/2026-08-16_periods/chartblock_month_2026-07.md \
--relpath figures/month_2026-07
python -m ad_ops.measures --config _runs/2026-08-16_receipt_series/run_config_full_meta.json \
--from 2026-07-23 --to 2026-07-31
上面最后一条命令是唯一一条其口径行构成有效校验的,因为它跑在账户自己的时区上。generate.sh 一次跑 完六个周期和每一个标签页;每个广告系列写进自己的 --out 目录,因为图表文件名里不带广告系列,两个 广告系列共用一个目录会互相覆盖。
11 附录 — 定义
报表周期是一串 UTC 日。 Adjust 按 UTC 报数,按 UTC 分桶的报告与归因面板天然吻合,不需要任何人 在脑子里换算。B1 固定在 UTC-7,因此这个账户上的一个 UTC 日等于账户日 D−1 的 17:00 到 23:59加上账户 日 D 的 00:00 到 16:59。
分小时一节把一天中的小时跨周期合并。 九天有 216 个小时,而表格只有 24 行,所以每一行是九个小时 的买量叠在一起,每天各出一个。它读的是一天之内的形状,不是时间轴,表里没有哪一行是一次事件。
总量在长度不同的周期之间不可比,比率可以。 本报告只印一个周期、完全不做对比,所以这里不存在这 个风险;但这条规则正是 8 月的报告要比 CPM、CTR、点击→安装、CPI、ROAS、付费率、ARPU 与 ARPPU 的原 因。
三种收入口径 ,从不只用一种,从不混用,永远标注:
口径 计什么 本期结束后还会动吗 入账收入 在本周期之内到账的钱,不论付费用户哪天安装 午夜后约四小时结清,此后固定。Meta 自己的数字就是这个口径 当日回收 当天安装并且 在当天结束前付费 永远不动。 午夜封存同期群 H 小时 当天的安装,各自从自己安装起算 H 小时内的付费。默认视野 H=24 在所有安装都活满 H 小时前持续上升,之后冻结。必须写出覆盖率
跨日对比用当日回收口径,因为它不可变且不需要视野,代价是它会按一天中的时点做截尾:01:00 安装的人在 午夜前有 23 小时可以付费,23:00 安装的人只有一小时。固定视野的同期群口径对时点是中性的。入账是 Meta 会拿来跟你说的数字,写出来是为了解释差额,不是为了排次序。三种口径都读在我们自己的付款记录上 ;Meta 只能表达入账,因此它跟在那一列旁边的括号里,永远不单独成行。
放到一个周期上,入账收入按付费用户已安装的天数拆分。 在一天之内,已安装天数与安装日是同一件事 ;到了九天,两者就分开了,这个拆分中安装当天那一行,恰好等于当日回收的合计。
存在两套安装数,它们从不交叉相除。 Meta 的 Leads 与我们自己的注册数相差 5% 到 10%。投放漏斗用 Meta 的,因此 CPI 与点击→安装的分子分母都在同一把尺子上。付费率与所有收入口径用我们自己的,因此付费用 户与他们来自的安装是同一批人。
预算区间 是广告组日预算没有变动的一段时间。它之所以重要,是因为预算在这个账户上是一个地域旋钮: 调高预算会买到更宽、更便宜的国家,跨过一次预算变动去比较两条素材,比的是两批不同的人。
16 付费用户门槛 (DECISION_LOG.md #37):付费用户数低于约 16 时,一个收入数字只能看方向,不能 看幅度。这个账户上每次安装的收入是一个几乎处处为零、偶有大额的向量,要把两条素材分出五分之一的差距 ,每条实验臂需要上千次安装。
这些广告系列的名字是什么意思。 T1、T2、T3 是账户成组买入的国家分组,由 DECISION_LOG.md #24 固定下来:T1 是美国、加拿大、英国、澳大利亚与新西兰;T2 是阿联酋、沙特、马来西亚、南非与墨西哥;T3 是印度、印尼、尼日利亚、巴基斯坦与菲律宾。它们一共跑过两轮:2026-07-23 那一轮,素材标签前缀里带 a;2026-07-25 那一轮,带 b。第一轮测的是两个文案方向,utility 讲把一张照片变成视频, control 讲一个按你的意思动的陪伴对象,各用一条十五秒的片子。第二轮把这两个方向乘上三种时长,在三 个分组里各跑一遍,十八个格、每格一个广告组。sasian_ 与 latina_es_ 是 2026-07-26 加进第二轮印度 与墨西哥广告组的南亚素材包与西语素材包;紧跟其后的词是片中的服装,数字是片长。旧广告系列是 2026-07-30 取代这套分组结构的全球买量,德语区广告系列是同一天开投的 DACH 测试。