當商品本身，就是最大的過審障礙
When the Product Itself Was the Obstacle
Meta 的審核機制持續判定商品照片裸露過多，過不了審。
Meta's review kept flagging the product photography for showing too much skin, so only single-image post ads reliably survived review.
應該扛起帳戶大半流量的 DPA(動態商品廣告)，幾乎每天被擋下來。Meta 的審核機制持續判定商品照片裸露過多，過不了審。我接手前，唯一能穩定過審的只剩單圖貼文廣告，因此帳戶失去了動態再行銷與商品目錄驅動的再行銷，實際被消費者看到的商品比例也一路萎縮。
Dynamic Product Ads, the format that should have carried most of the account's weight, were getting rejected almost every day. Meta's review kept flagging the product photography for showing too much skin, so only single-image post ads reliably survived review. That cost the account its dynamic retargeting and catalog-driven remarketing, and shrank how much of the catalog actually reached shoppers.
目錄健康度恢復之後，DPA 才真正跑得起來，審核被擋的狀況也從此大幅減少。
With the feed healthy again and the primary images swapped out, the DPA campaigns could finally run, and rejections dropped sharply.
我重新調整了商品目錄架構與分類，排除掉少數幾乎不可能過審的品類，再換掉驅動 DPA 的商品首圖。目錄健康度恢復之後，DPA 才真正跑得起來，審核被擋的狀況也從此大幅減少。受眾端前期每週微調，測出穩定架構後就不再動。出價每週檢視，只有在市場競爭激烈的檔期才調整。素材跟著活動節奏，每週上稿兩到三次。
I rebuilt the product catalog structure and its category groupings, and pulled the handful of product lines that were never realistically going to pass review. With the feed healthy again and the primary images swapped out, the DPA campaigns could finally run, and rejections dropped sharply. On the audience side, I adjusted targeting weekly at first, then left the structure alone once it proved stable. Bids got a weekly look and moved only during competitive periods. Creative went up two or three times a week, timed to the campaign calendar.
ROAS 在兩到三個月內從 1.5 拉到 9，並維持到合作結束。
ROAS went from 1.5 to 9 within two to three months and held there for the rest of the engagement.
ROAS 在兩到三個月內從 1.5 拉到 9，並維持到合作結束。這個數字的組成要講清楚:它是 Meta 加 Google(含品牌關鍵字)的綜合結果。去掉品牌字，純廣告曝光帶來的 ROAS 大約落在 4 到 5，這個數字才真正說明帳戶有沒有能力找到新客，而不只是收割已經認識品牌的人。全站流量成長 50%，收益成長 95%，會員成長 80%。
ROAS went from 1.5 to 9 within two to three months and held there for the rest of the engagement. It's worth being precise about what's inside that number: it's the blended result across Meta and Google, including branded search. Strip out brand terms and the pure prospecting ROAS still lands around 4 to 5, which is the number that actually shows whether the account can find new customers rather than just closing out people who already knew the brand. Site-wide traffic grew 50%, revenue grew 95%, and membership signups grew 80%.
先修真正壞掉的地方，即使解法不是廣告投放本身。
One decision from that engagement sat outside media buying entirely.
這次合作有一個決定完全在投放範疇之外:品牌當時毛利是 -15%，一個月內我透過調整定價與營運策略把它收斂到 -7%，同時接受營業額掉了將近 20% 這個代價。這跟我做事的邏輯是同一套:先修真正壞掉的地方，即使解法不是廣告投放本身。
One decision from that engagement sat outside media buying entirely. The brand was running a negative margin of -15%. Within a month of restructuring pricing and operations, I brought that to -7%, accepting a nearly 20% drop in revenue as the deliberate cost of that trade. It's the kind of call I'd make again: fix what's actually broken, even when the fix has nothing to do with the media plan.