常見問題總覽 Frequently Asked Questions

看的是總成本，不是單一報價。代理商的「低顧問費」通常由媒體佣金與價差補貼——你多花的廣告費裡有一部分其實是隱藏費用，而且它讓建議偏向「多投放」。零佣金顧問的報價就是全部成本，建議不被媒體收入影響。比較兩者時，應該把「顧問費＋因偏誤建議多花的媒體費」一起算。 Compare total cost, not the sticker price. An agency's "low retainer" is usually subsidized by media commissions and spreads — part of your ad spend is a hidden fee, and it biases advice toward "spend more." A zero-commission quote is the entire cost, and the advice isn't shaped by media revenue. The fair comparison is: advisory fee + the extra media spend that biased advice causes.

GROVA 的定位不是取代代理商，而是站在你這邊看代理商的成效。代理商負責執行投放，GROVA 負責驗證哪些投放真的帶來增量、幫你設定對代理商的 KPI 與談判依據。多數客戶在引入獨立視角後，代理商的產出品質反而提升——因為有人在替你認真看報表。 GROVA doesn't replace your agency — we sit on your side of the table and evaluate its output. The agency executes; GROVA verifies which spend actually drives incremental revenue, and arms you with KPIs and negotiation grounds. Most clients find agency quality improves once someone is genuinely reading the reports for you.

不會。GROVA 不代操廣告、不代理媒體——這是零佣金模式的前提。我們負責診斷、策略與驗證：告訴你錢該往哪裡配、怎麼驗證有沒有效；實際投放由你的內部團隊或代理商執行。這個分工讓我們的建議永遠不受媒體收入影響。 No. GROVA doesn't operate ads or sell media — that's the premise of the zero-commission model. We diagnose, strategize, and verify: where the money should go and how to prove it worked. Execution stays with your team or agency. That separation is what keeps our advice free of media revenue bias.

每一項建議交付時都附帶 ROI 假設與驗證方法，所以「有沒有效」不是感覺問題，是可以複驗的。如果驗證結果與假設不符，我們會公開檢討是假設錯、資料錯還是執行走樣——方法論全程透明，你的團隊看得到每一步怎麼算的。我們不承諾保證成效（會這樣承諾的顧問值得警惕），但承諾每個結論都可以被檢驗。 Every recommendation ships with an ROI hypothesis and a verification method — "did it work" is testable, not a feeling. If results diverge from the hypothesis, we openly diagnose whether the assumption, the data, or the execution was off. We don't guarantee outcomes (be wary of consultants who do); we guarantee every conclusion can be checked.

簽。開案前即可簽署 NDA，涵蓋你提供的所有營運數據、媒體花費與策略資訊。同樣的保密標準也適用於我們過去的合作對象——這也是為什麼網站上的實績案例一律匿名化，只呈現產業大類與數字。 Yes — before the engagement starts, covering all operational data, media spend, and strategic information you share. The same standard applies to our past work, which is why every track-record item on this site is anonymized to industry level and numbers only.

因為每個平台都把同一筆訂單記到自己頭上。消費者下單前可能看過 Meta 廣告、點過 Google 搜尋、又收過 LINE 推播——三個平台各記一次轉換，加總自然超過實際營收，常見是兩到三倍。這不是造假，是歸因邏輯的本質限制。解法是用平台之外的共同框架（例如 MMM）衡量增量貢獻，而不是讓每個平台自己給自己打分數。 Because every platform credits itself for the same order. A customer might see a Meta ad, click a Google search result, and get a LINE push before buying — three platforms each log a conversion, so the sum runs 2–3× actual revenue. It's not fraud; it's the nature of attribution. The fix is a platform-independent framework like MMM — not letting each platform grade its own homework.

依序排除五層：一、資料層——先確認追蹤沒壞，數字是真的掉；二、流量層——量掉了還是結構變了；三、轉化層——漏斗哪一步的轉換率下降；四、客單層——單價或購買件數變化；五、結構層——品類組合、季節性或競品動作。多數品牌一路跳到「加廣告預算」，但問題常在第一層或第三層。順序錯了，錢就白花。 Rule out five layers in order: (1) Data — is tracking broken, or is the drop real; (2) Traffic — volume down, or mix shifted; (3) Conversion — which funnel step declined; (4) Basket — price or units per order; (5) Structure — category mix, seasonality, competitor moves. Most brands jump straight to "increase ad budget," but the problem usually sits in layer 1 or 3. Wrong order, wasted money.

沒有跨品類通用的健康值——消耗品和耐久財的合理回購率差距可以是十倍。比單一數字更有用的是三個問題：一、你的回購率趨勢是升是降；二、同品類競品大約在哪個區間；三、第二次購買的間隔是否在縮短。如果首購後 90 天內回購率明顯低於品類常態，通常代表 CRM 序列有缺口，這是最容易先修的地方。 There's no universal benchmark — consumables and durables can differ by 10×. Three better questions: Is your rate trending up or down? Where does your category roughly sit? Is the gap between first and second purchase shrinking? If 90-day repurchase is clearly below category norms, your CRM sequences probably have a gap — and that's the cheapest thing to fix first.

頻率不是重點，相關性才是——封鎖率才是你該盯的指標。原則有三：一、分眾後再推，全名單廣播是封鎖率最高的做法；二、依會員生命週期設計觸發（首購後、回購週期到期前、沉睡喚醒），而不是照行事曆亂槍打鳥；三、每次推播都要能回答「這則訊息對這群人現在有什麼用」。做到分眾觸發的品牌，週推一到兩次通常安全；做不到的，月推一次也嫌多。 Frequency isn't the variable — relevance is, and block rate is the metric to watch. Three rules: segment before you send (full-list blasts have the highest block rates); trigger by lifecycle (post-first-purchase, pre-repurchase-window, win-back) instead of by calendar; and every push must answer "why is this useful to this group right now." With segmented triggers, 1–2 pushes a week is usually safe; without them, once a month is already too much.

不要用「市場大小」選，要用「驗證成本」選。新加坡市場小但法規透明、物流成熟、英語溝通，適合當測試場；馬來西亞有華語滲透與較低的進入成本，適合驗證華人圈以外的接受度；直接進印尼、越南等大市場的失敗率最高，因為在地化門檻被普遍低估。正確的問題不是「哪個市場最大」，而是「哪個市場能用最低成本告訴我產品到底行不行」。 Choose by validation cost, not market size. Singapore is small but transparent, logistically mature, and English-speaking — a good test bed. Malaysia offers Chinese-language penetration and lower entry costs — good for testing beyond the Chinese-speaking circle. Going straight into Indonesia or Vietnam fails most often, because localization barriers are chronically underestimated. The right question isn't "which market is biggest" but "which market tells me, at the lowest cost, whether my product actually works."

設計得當的返利重設，是讓「有做事的經銷商賺更多」，不是全面砍利潤。做法上先用數據分層：哪些經銷商真的在推你的產品、哪些只是搬貨賺價差；新結構把獎勵綁到你要的行為（鋪貨、陳列、新客開發），並保留過渡期。會反彈的通常是原本靠資訊不對稱獲利的少數，而他們本來就不是你成長的來源。 A well-designed rebate reset means "distributors who do the work earn more" — not an across-the-board margin cut. Start with data: which partners actually push your product versus just arbitraging price spreads. The new structure ties rewards to behaviors you want (distribution, display, new account development), with a transition period. Pushback usually comes from the few who profited on information asymmetry — and they were never your growth engine anyway.

AEO（Answer Engine Optimization，答案引擎優化）是讓內容被 ChatGPT、Perplexity、Google AI Overviews 等 AI 系統理解、抽取並引用為答案來源的優化方法。SEO 優化的是「排名」——讓人點進你的網站；AEO 優化的是「被引用」——讓 AI 在回答問題時採用你的定義與觀點。兩者的內容基本功重疊（結構清楚、意圖明確、有專業可信度），但 AEO 更要求：結論先行、定義可獨立抽取、FAQ 與結構化資料完整。 AEO (Answer Engine Optimization) makes content understandable, extractable, and citable by AI systems like ChatGPT, Perplexity, and Google AI Overviews. SEO optimizes for rankings — getting people to click through; AEO optimizes for citations — getting AI to adopt your definitions and viewpoints when it answers. The fundamentals overlap (clear structure, clear intent, credibility), but AEO additionally demands conclusion-first writing, standalone extractable definitions, and complete FAQ and structured data.

五個實作重點：一、每頁開頭放 40–80 字可獨立成立的結論或定義，AI 抽取的就是這種段落；二、用問句當標題，標題下第一句直接回答；三、建立清楚的組織實體（Organization schema、一致的公司描述），AI 引用的是「可辨識的來源」；四、FAQ 與比較表是被引用率最高的格式；五、內容要有明確主張與數字級距，空泛的品牌文案沒有引用價值。 Five practical moves: (1) open every page with a 40–80 character conclusion or definition that stands alone — that's what AI extracts; (2) use questions as headings, with the answer in the first sentence below; (3) build a recognizable organization entity (Organization schema, a consistent company description) — AI cites identifiable sources; (4) FAQs and comparison tables get cited most; (5) make concrete claims with number ranges — vague brand copy has zero citation value.

對資訊型查詢，會——用戶在搜尋結果頁就拿到答案，點擊率下降是既定趨勢。但對高價 B2B 服務來說，真正重要的流量本來就不是資訊型，而是「比較與決策型」：這類查詢 AI 給出摘要後，用戶仍會點進來源驗證。策略上應該把資訊型內容當「被引用的入口」（建立權威），把轉換寄託在決策型頁面（服務頁、比較頁、FAQ）。流量會變少，但進站的人意圖更強。 For informational queries, yes — users get the answer on the results page, and click-through decline is a settled trend. But for high-value B2B services, the traffic that matters was never informational; it's comparison and decision queries, where users still click through to verify sources. Strategically: treat informational content as your citation gateway (authority building), and put conversion weight on decision pages — services, comparisons, FAQs. Less traffic, higher intent.

Google 的官方立場是看品質不看產製方式——懲罰的是「為操縱排名而大量生產的低價值內容」，不是 AI 本身。實務上的分界線：有沒有第一手經驗與具體判斷（E-E-A-T）、有沒有明確主張與可驗證資訊、是不是解決真實搜尋意圖。用 AI 起草、由專業者注入實戰觀點與數字的內容沒有問題；純 AI 洗稿、無觀點的內容農場文，才是風險所在。 Google's stated position is that quality matters, not how content was produced — what gets penalized is low-value content mass-produced to manipulate rankings, not AI itself. The practical dividing line: first-hand experience and concrete judgment (E-E-A-T), clear claims with verifiable information, and genuine search intent. AI-drafted content with expert viewpoints and real numbers injected is fine; zero-viewpoint AI content-farm output is where the risk lives.

AI 取代的是不透明，不是判斷。過去顧問的部分價值來自資訊不對稱——框架、模板、產業報告，這些 AI 都能給你。留下來的價值有三種 AI 給不了：一、對你的資料做過驗證的判斷（AI 不知道你的追蹤壞了）；二、利益獨立的立場（AI 不會幫你跟代理商談判）；三、對結論負責（建議錯了有人要檢討）。所以問題不是要不要顧問，而是顧問的方法論透不透明——黑箱顧問會被 AI 淘汰，方法公開、可驗證的顧問會被 AI 放大。 AI replaces opacity, not judgment. Part of consulting's old value was information asymmetry — frameworks, templates, industry reports — and AI now gives you all of that. What remains is what AI can't provide: judgment validated against your actual data (AI doesn't know your tracking is broken), an independent position (AI won't negotiate with your agency), and accountability for conclusions (someone answers when the advice is wrong). The question isn't whether to hire a consultant — it's whether their methodology is transparent. Black-box consultants get replaced by AI; open, verifiable ones get amplified by it.