執行腳本：python3 api/daiken_sd_amc_pause.py --execute（18 個活動已於 2026-04-20 執行，18 個活動全成功）

📄 下載 SnS Insights PDF 📊 下載 SnS Insights Excel

ASIN	Product	Units	Sales	活躍訂閱	滲透率	賣家出資
B0DTT8T7X1	Mini Fish Oil 2pk	54	$2,576	23	44.89%	10%
B0DTYCNVQV	Nattokinase	92	$2,542	57	17.12%	15%
B0DV3VX26C	Bitter Melon	60	$1,613	35	19.24%	10%
B09Y8VBQTV	Maca Powder	63	$1,447	72	7.24%	10%
B0DV3SJ8LK	Bitter Melon v2	22	$1,160	15	28.39%	10%
B0DTYLYN6B	Nattokinase v2	21	$1,101	20	15.36%	15%
B0G7K4CHWG	Kids Fish Oil Gummy	63	$1,054	27	7.80%	5%
B0G7K1VC4G	Kids Fish Oil (child)	30	$986	26	18.53%	10%
B0DTTHJCRL	Mini Fish Oil 60pk	24	$638	16	18.72%	10%
B0FH6DDW63	Variant	11	$472	5	35.09%	10%
B0DTTT9WTN	Mini Fish Oil (child)	12	$221	10	41.51%	10%
B0FH6BX4GQ	Variant	7	$168	2	13.93%	10%
B0DTTVGD2F	Mini Fish Oil (child)	6	$83	3	21.99%	10%
	合計 (13 ASINs)	464	$14,061	311

🎯 關鍵對照（SD AMC 歸因 vs Seller Central 實際）
納豆 Nattokinase（B0DTYCNVQV + B0DTYLYN6B）：77 活躍訂閱戶、$3,643 SnS 90d 銷售 — 完全沒開廣告，純 organic 訂閱 → 驗證「自然單撐」策略正確
Maca：72 活躍訂閱戶（最多），但滲透率 7.24%（最低）→ 代表 Maca 買家多數沒設 SnS，是 SnS_Retarget_Maca 的理想目標。但 14 天 $52 花 0 歸因訂閱 = 沒打中
Mini Fish Oil 2pk：滲透率 44.89%（最高）— 既有買家訂閱忠誠度極強，不需再推
Buy1_NoSubscribe 目標群：Kids Fish Oil Gummy (B0G7K4CHWG) 滲透率只 7.80% — 大量未訂閱買家；但 14 天廣告只換 1 單

資料來源：Seller Central → Reports → Subscribe & Save Performance（2026-01-18 ~ 2026-04-18）· 原始檔：clients/daiken/input/sns_90d_20260118_20260418.json

-- 主查詢（Part 1）— AMC SQL (SECONDS_BETWEEN only, no DAY/INTERVAL)
-- 14 days = 14 × 86400 = 1,209,600 seconds
WITH campaign_exposure AS (
  SELECT campaign, campaign_id, user_id, MIN(event_dt_utc) AS first_exposure_dt
  FROM display_traffic
  WHERE campaign IN (
    'DAIKEN SD AMC_Buy1_NoSubscribe_20260410',
    'DAIKEN SD AMC_SnS_Retarget_FishOil_20260330',
    'DAIKEN SD AMC_SnS_Retarget_Maca_20260330',
    'DAIKEN SD AMC_SnS_Retarget_BitterMelon_20260330'
  )
  GROUP BY 1, 2, 3
),
attributed_subs AS (
  SELECT e.campaign, e.user_id, s.event_dt_utc AS sns_dt, s.conversions,
         SECONDS_BETWEEN(e.first_exposure_dt, s.event_dt_utc) / 86400.0 AS days_to_sub
  FROM campaign_exposure e
  INNER JOIN conversions_all_for_audiences s ON e.user_id = s.user_id
  WHERE s.event_type_description = 'New SnS Subscription'
    AND SECONDS_BETWEEN(e.first_exposure_dt, s.event_dt_utc) >= 0
    AND SECONDS_BETWEEN(e.first_exposure_dt, s.event_dt_utc) <= 1209600
)
SELECT campaign, COUNT(DISTINCT user_id) AS unique_subscribers,
       SUM(conversions) AS total_new_sns_events, AVG(days_to_sub) AS avg_days_to_subscribe
FROM attributed_subs
GROUP BY 1 ORDER BY 2 DESC;

🎯 預期結果判讀
如 AMC 回傳 unique_subscribers = 0 for all 4 campaigns — 證明 SnS Retarget + Buy1_NoSubscribe 沒把人變訂閱戶，昨天暫停決策正確。
如果有數字（例如 Buy1 = 2 subs）— 那 $147 花 14d 換 2 訂閱，每 subs 獲客成本 $73（vs 訂閱 LTV 需算）