Back to Details

amz-ppc-campaign

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

PPC Campaign Builder

PPC广告活动构建指南

Sponsored Products is the engine of most Amazon ad accounts. Built right, it is a funnel that discovers keywords and then harvests them profitably. Built wrong, it is one giant auto campaign quietly losing money. This skill builds the funnel or audits the one you have, and adds dayparting as an advanced layer once the structure is sound and the data is thick enough to justify it.
Sponsored Products是大多数亚马逊广告账户的核心引擎。搭建得当的话，它是一个既能挖掘关键词又能实现盈利的漏斗模型；搭建不当的话，它就只是一个默默亏损的大型自动广告活动。本技能可帮你搭建该漏斗模型，或审计现有模型，并在结构合理、数据足够充足的基础上，添加dayparting（分时投放）这一进阶优化层。

When to use this

适用场景

  • A new product is launching and needs a Sponsored Products structure from scratch.
  • An existing PPC account has no funnel, just one auto campaign that never scales.
  • ACoS is high and a search term report is sitting unread.
  • Bids were set by guess and the seller wants them tied to margin.
  • A seller wants a repeatable structure across multiple SKUs.
  • 新品上线，需要从零搭建Sponsored Products广告活动结构
  • 现有PPC账户无漏斗模型，仅靠一个自动广告活动无法实现规模增长
  • ACoS偏高，且搜索词报告尚未被分析利用
  • 出价全凭猜测，卖家希望将出价与利润挂钩
  • 卖家希望为多个SKU搭建可复用的广告活动结构

Two modes

两种模式

  • Mode A. Build. A new product, no campaigns yet. Output: a campaign structure.
  • Mode B. Audit. Campaigns exist. Output: bid changes, negatives, graduations.
  • 模式A：搭建。适用于新品、尚未创建广告活动的场景。输出：完整的广告活动结构。
  • 模式B：审计。适用于已有广告活动的场景。输出：出价调整建议、否定关键词、关键词升级方案。

The framework. The Discovery to Harvest Funnel

框架：从发现到收割的漏斗模型

Sponsored Products works as a three-campaign funnel. keywords flow left to right as they prove themselves.
DISCOVER            TEST               HARVEST
Auto campaign  -->  Broad/phrase   -->  Exact-match
finds terms         campaign            campaign
                    confirms intent     scales the winners
  • Discover. An auto campaign. Amazon matches your product to search terms you did not think of. The search term report is the gold.
  • Test. Promising terms from Discover go into a broad or phrase campaign to confirm they convert with real intent.
  • Harvest. Terms that convert in Test graduate to an exact-match campaign with higher bids. this is the profit engine. As a term graduates, it is added as a negative in the campaign it came from, so the funnel does not bid against itself.
Sponsored Products的运作遵循三阶段广告漏斗模型，关键词会随着表现提升从左向右流动。
DISCOVER            TEST               HARVEST
Auto campaign  -->  Broad/phrase   -->  Exact-match
finds terms         campaign            campaign
                    confirms intent     scales the winners
  • 发现阶段：自动广告活动。亚马逊会将你的产品匹配到你未曾想到的搜索词，搜索词报告是核心价值所在。
  • 测试阶段：将发现阶段中有潜力的词加入广泛/词组匹配广告活动，验证其真实转化意向。
  • 收割阶段：在测试阶段转化表现良好的词会升级到精准匹配广告活动，并设置更高出价，这是盈利的核心引擎。当关键词升级后，需将其添加为来源广告活动的否定关键词，避免漏斗内部相互竞价。

The bid math

出价计算逻辑

Every bid traces back to margin.
  • Break-even ACoS equals contribution margin divided by price. the most an ad can cost before the unit loses money.
  • Target ACoS is set below break-even for Harvest (you want profit) and may be at or above break-even for Discover (you are buying keyword data).
  • Max CPC roughly equals price, times target ACoS, times expected conversion rate. starting bids come from this, not from a guess.
所有出价均基于利润推导而来。
  • 盈亏平衡ACoS = 边际利润 ÷ 产品售价，这是广告成本的上限，超过该值单品会亏损。
  • 目标ACoS：收割阶段需设置在盈亏平衡值以下（追求盈利），发现阶段可设置为等于或高于盈亏平衡值（用于获取关键词数据）。
  • 最高CPC ≈ 产品售价 × 目标ACoS × 预期转化率。初始出价由此公式计算，而非凭空猜测。

Step by step. Mode A, Build

分步操作：模式A（搭建）

  1. Collect price, contribution margin, the keyword set (or run keyword logic), and the product phase.
  2. Compute break-even ACoS, target ACoS per funnel stage, and starting Max CPC.
  3. Build the three campaigns: one auto for Discover, one broad or phrase for Test, one exact for Harvest, seeded with the strongest known keywords.
  4. Group keywords and set match types and starting bids from the bid math.
  5. Define the negative-keyword isolation so the three campaigns do not compete.
  1. 收集产品售价、边际利润、关键词组（或执行关键词挖掘逻辑）以及产品所处阶段。
  2. 计算盈亏平衡ACoS、各漏斗阶段的目标ACoS，以及初始最高CPC。
  3. 创建三个广告活动：一个用于发现阶段的自动广告、一个用于测试阶段的广泛/词组匹配广告、一个用于收割阶段的精准匹配广告，并植入最优质的已知关键词。
  4. 对关键词进行分组，根据出价计算逻辑设置匹配类型和初始出价。
  5. 定义否定关键词隔离规则，避免三个广告活动相互竞争。

Step by step. Mode B, Audit

分步操作：模式B（审计）

  1. Collect the search term report and campaign data: spend, clicks, orders, sales, ACoS per term, plus target ACoS.
  2. Set the bid action for each term that sells: raise bid (converts below target ACoS), or lower bid (converts above target ACoS but still sells).
  3. Sort the rest into negate, keep-and-graduate, or wait. For the full negate/keep/ graduate decision tree (the drain threshold, exact versus phrase negatives, and protecting converting terms from a careless phrase negative), see amz-negative-keywords.
  4. Apply negative isolation for every graduated term, so the funnel does not bid against itself.
  5. Produce a week-by-week action plan.
  1. 收集搜索词报告和广告活动数据：花费、点击量、订单量、销售额、各关键词的ACoS，以及目标ACoS。
  2. 为有转化的关键词设置出价操作：若转化后ACoS低于目标则提高出价，若转化后ACoS高于目标但仍有销量则降低出价。
  3. 将剩余关键词分为三类：添加否定、保留并升级、观察等待。如需完整的否定/保留/升级决策树（包括淘汰阈值、精准与词组否定的区别、避免误否定有转化的关键词），可查看amz-negative-keywords。
  4. 为所有升级的关键词应用否定隔离规则，避免漏斗内部相互竞价。
  5. 制定每周行动计划。

Output format

输出格式

undefined
undefined

PPC Plan. [product] . Mode [A/B]

PPC计划. [产品名称] . 模式[A/B]

Break-even ACoS: [%] Target ACoS: Harvest [%] / Discover [%]
盈亏平衡ACoS: [%] 目标ACoS: 收割阶段[%] / 发现阶段[%]

Mode A. Structure

模式A：结构

Discover (auto): [budget, starting bid] Test (broad/phrase): [keyword groups, bids] Harvest (exact): [seed keywords, bids] Negative isolation: [rules]
发现阶段（自动）: [预算, 初始出价] 测试阶段（广泛/词组）: [关键词组, 出价] 收割阶段（精准）: [种子关键词, 出价] 否定隔离规则: [规则内容]

Mode B. Actions

模式B：操作建议

Raise bid: [terms, new bids] Lower bid: [terms, new bids] Negate: [terms] Graduate to exact: [terms] Week-by-week plan: ...
undefined
提高出价: [关键词, 新出价] 降低出价: [关键词, 新出价] 添加否定: [关键词] 升级至精准匹配: [关键词] 每周行动计划: ...
undefined

Worked example

示例演示

Product 30 USD, contribution margin 12. Break-even ACoS 40 percent.
Mode A: target ACoS 25 percent for Harvest, 50 percent allowed for Discover. Three campaigns. auto for discovery, broad for testing, exact for the 8 strongest known keywords with bids set from a Max CPC of roughly 30 x 0.25 x conversion rate. Each exact keyword is negated in the auto and broad campaigns so the funnel does not bid against itself.
产品售价30美元，边际利润12美元。盈亏平衡ACoS为40%。
模式A：收割阶段目标ACoS设为25%，发现阶段允许设为50%。创建三个广告活动：自动广告用于发现，广泛匹配广告用于测试，精准匹配广告植入8个最优质的已知关键词，出价基于最高CPC公式（约30×0.25×转化率）计算。每个精准关键词需在自动和广泛广告活动中设置为否定关键词，避免漏斗内部相互竞价。

Quality check

质量核查

  • Break-even ACoS is computed from margin, and target ACoS differs by funnel stage.
  • The structure is a Discover, Test, Harvest funnel, not one campaign.
  • Every graduated keyword is negated in its source campaign. no internal competition.
  • Bids trace to the Max CPC math, not a flat guess.
  • Mode B decisions distinguish raise, lower, negate, and wait. low-click terms wait.
  • 盈亏平衡ACoS由利润计算得出，且各漏斗阶段的目标ACoS不同
  • 结构为发现-测试-收割的漏斗模型，而非单一广告活动
  • 所有升级的关键词均在来源广告活动中设置为否定关键词，无内部竞争
  • 出价基于最高CPC公式计算，而非统一出价
  • 模式B的决策需区分提高、降低、否定、观察四种操作，低点击量关键词需观察等待

Common mistakes

常见误区

  • One giant auto campaign. Discovery with no harvest. it never scales the winners.
  • No negative isolation. The auto, broad, and exact campaigns bid against each other for the same term and inflate the cost.
  • Flat bids. One bid across all keywords ignores that they convert differently.
  • Never graduating. Winners left in the broad campaign forever instead of moving to a higher-bid exact campaign.
  • Judging Discover by Harvest ACoS. Discovery buys keyword data. it is allowed to run hotter.
  • 单一大型自动广告活动：仅做发现阶段而无收割阶段，无法放大优质关键词的效果
  • 无否定隔离规则：自动、广泛、精准广告活动为同一关键词相互竞价，推高成本
  • 统一出价：所有关键词使用同一出价，忽略不同关键词的转化差异
  • 从不升级关键词：优质关键词一直留在广泛广告活动中，未转移至高出价的精准广告活动
  • 用收割阶段的ACoS评判发现阶段：发现阶段的目的是获取关键词数据，允许更高的ACoS

Advanced. Dayparting

进阶内容：Dayparting（分时投放）

Once the funnel is built and a healthy data stream exists, the next lever is dayparting. shifting bids and budgets toward the hours that convert and away from the hours that drain. This is an optimization on top of the funnel, not a substitute for it.
当漏斗模型搭建完成且数据稳定后，下一个优化方向是dayparting（分时投放）——将出价和预算向高转化时段倾斜，远离低转化时段。这是在漏斗模型基础上的优化，而非替代方案。

Data-quality gate

数据质量门槛

Before doing any of this, the account must clear two bars. At least 4 weeks of hour-and-day performance data and enough daily spend that hour-level slices have meaningful click volume (rule of thumb. if a typical hour has under 5 clicks across the data window, treat it as noise). Below that, dayparting is acting on noise. skip it, fix structure and bids first, come back when the data is thick enough.
If only day-of-week data is reliable, daypart by day, not hour. group thin hours into wider windows (morning, afternoon, evening, overnight) until each window has enough clicks to trust.
在进行分时投放前，账户需满足两个条件：至少4周的小时级+日期级表现数据，以及每日花费足够高，使得小时级数据具备有意义的点击量（经验法则：若某时段在数据周期内的平均点击量不足5次，则视为无效数据）。未达标的话，分时投放是基于无效数据的操作，应先优化结构和出价，待数据充足后再进行。
若仅日期级数据可靠，则按日期而非小时进行分时投放；将数据不足的小时合并为更宽的时段（上午、下午、晚上、夜间），直到每个时段的点击量足够可信。

The Hour Value Map

时段价值地图

Dayparting is built from the account's own hour-by-day conversion, not from intuition.
  1. Score every time block. For each hour, or each hour-and-weekday block, compute conversion rate and ACoS. Rank blocks into three tiers:
    • Prime. Conversion above the daily average and ACoS at or below target.
    • Standard. Around the daily average.
    • Drain. Conversion well below average or ACoS far above target.
  2. Assign a bid posture per tier.
    • Prime: bid up, plus 15 to 30 percent. this is where budget should land.
    • Standard: baseline bid.
    • Drain: bid down 30 to 50 percent, or pause entirely.
  3. Protect the budget for Prime. The point of cutting Drain hours is so the daily budget survives until the Prime hours arrive. A budget that empties at noon never reaches the evening peak. reallocate, do not just cut.
分时投放需基于账户自身的时段转化数据，而非主观判断。
  1. 为每个时段打分：针对每个小时（或小时+工作日组合），计算转化率和ACoS，将时段分为三个等级：
    • 黄金时段：转化率高于日均水平，且ACoS等于或低于目标值
    • 标准时段：转化率接近日均水平
    • 低效时段：转化率远低于日均水平，或ACoS远高于目标值
  2. 为每个等级设置出价策略：
    • 黄金时段：提高出价15%-30%，预算应优先分配至该时段
    • 标准时段：使用基准出价
    • 低效时段：降低出价30%-50%，或直接暂停投放
  3. 保障黄金时段的预算：削减低效时段预算的目的是让每日预算能维持到黄金时段。若预算在中午就耗尽，就无法覆盖晚间高峰。需重新分配预算，而非单纯削减。

Dayparting step by step

分时投放分步操作

  1. Confirm the data-quality gate (4+ weeks, sufficient hourly clicks).
  2. Score every block (or window) into Prime, Standard, Drain.
  3. Build the bid-adjustment schedule.
  4. Reallocate the freed budget into Prime. net spend can stay flat while results improve.
  5. Set a monthly re-score cadence. dayparting drifts with seasons and promotions.
  1. 确认满足数据质量门槛（4周以上数据、小时级点击量充足）
  2. 将每个时段（或合并后的时段）划分为黄金、标准、低效等级
  3. 制定出价调整时间表
  4. 将低效时段释放的预算重新分配至黄金时段，总花费可保持不变，同时提升效果
  5. 设置每月重新打分的节奏，分时投放策略会随季节和促销活动变化

Dayparting worked example

分时投放示例演示

A storefront product with 6 weeks of data. Overnight 0 to 6 AM converts at one third the daily average with ACoS double the target. Evening 6 to 10 PM converts well above average at target ACoS. Schedule: pause or cut overnight 40 percent. raise evening bids 25 percent. The daily budget no longer drains before the evening peak. net spend roughly flat, orders up, ACoS down.
某店铺产品拥有6周的数据。夜间0-6点的转化率仅为日均水平的1/3，ACoS是目标值的两倍；晚间6-10点的转化率远高于日均水平，且ACoS符合目标。策略：夜间时段暂停投放或降低出价40%，晚间时段提高出价25%。每日预算不再在晚间高峰前耗尽，总花费基本不变，订单量提升，ACoS下降。

Dayparting common mistakes

分时投放常见误区

  • Dayparting too early. Acting on a few days of data, or on hour buckets with 3 clicks each. that is hunch-driven, not data-driven.
  • Dayparting on a hunch. "Everyone shops at night" is not this account's data.
  • Cutting without reallocating. Cutting Drain hours and pocketing the budget, instead of moving it to Prime where it earns more.
  • Set and forget. A schedule built in Q1 is wrong by Q4. re-score monthly.
  • Ignoring day of week. Weekends and weekdays often convert differently. a day-of-week layer matters as much as the hour layer.
  • 过早进行分时投放：基于几天的数据或点击量仅3次的时段进行操作，这是主观判断而非数据驱动
  • 凭主观判断进行分时投放：“大家都在夜间购物”并非该账户的真实数据
  • 只削减不重新分配：削减低效时段预算后直接留存，未转移至黄金时段以获取更多收益
  • 设置后一成不变：第一季度制定的策略到第四季度已不再适用，需每月重新打分
  • 忽略日期差异：周末和工作日的转化表现通常不同，日期维度的分层和小时维度同样重要

Built by Jay GPT Pro

由Jay GPT Pro开发

Part of Amazon Pro Skills. Production-grade skills for serious Amazon sellers. Free and open. Built by Jay Margaliot.
I share a new AI play for Amazon sellers every week, free, in my WhatsApp group. Join here: https://chat.whatsapp.com/ILX65p1yWcaIG3c9WGHpTY
属于Amazon Pro Skills系列，为专业亚马逊卖家打造的实用技能。免费开源，由Jay Margaliot开发。
我每周会在WhatsApp群组中分享亚马逊卖家专属的AI玩法，免费加入：https://chat.whatsapp.com/ILX65p1yWcaIG3c9WGHpTY