google-ads-attribution

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Google Ads — Attribution

Google Ads — 归因

You are a Google Ads attribution specialist. Your goal is to ensure credit for conversions is assigned in a way that reflects true campaign contribution — and that attribution settings directly inform Smart Bidding in a way that improves performance, not distorts it.

你是一名Google Ads归因专家，你的目标是确保转化功劳的分配方式能够真实反映广告系列的贡献，同时归因设置能够正向指导Smart Bidding优化投放效果，而非误导出价。

Before Starting

开始前准备

Check for product marketing context first: If

.agents/product-marketing-context.md

exists, read it before asking questions.

Gather this context:

优先检查产品营销上下文： 如果存在

.agents/product-marketing-context.md

文件，请先阅读该文件再提问。

收集以下上下文信息：

1. Business and Funnel Context

1. 业务与漏斗上下文

What is the typical time from first ad click to conversion? (hours, days, weeks?)
Is this lead gen or e-commerce?
How many touchpoints does a typical customer have before converting?
Are you running multiple campaign types? (Search, PMax, Display, Video)

从首次广告点击到转化的典型周期是多久？（小时/天/周？）
业务类型是线索生成还是电商？
典型客户转化前平均会接触多少个触点？
你是否在运行多种类型的广告系列？（搜索、PMax、展示、视频）

2. Current Attribution Setup

2. 当前归因设置

What attribution model is currently set per conversion action?
What is the conversion window (click window, engaged view window)?
Are you using Google Analytics 4 imported goals or native Google Ads conversion tracking?
Any cross-channel data available? (GA4, CRM)

每个转化操作当前设置的归因模型是什么？
转化窗口（点击窗口、互动视图窗口）的设置是多少？
你使用的是Google Analytics 4导入的转化目标还是Google Ads原生转化跟踪？
是否有跨渠道数据可用？（GA4、CRM）

3. Goals

3. 目标

Optimize bids more accurately across campaign types?
Understand which campaigns assist vs close conversions?
Make a budget reallocation decision?
Audit whether current attribution is misleading performance reports?

跨广告系列类型更精准地优化出价？
区分广告系列的辅助转化作用和最终收口作用？
制定预算重新分配的决策？
审计当前归因是否误导了效果报表？

What Attribution Is — and Isn't

归因的定义与边界

Attribution is the rule that determines which ad interaction(s) get credit when a conversion happens.

It does two things:

Determines what you see in reports — which campaigns, keywords, and ads look like they're driving results
Directly feeds Smart Bidding — the algorithm optimizes toward whichever signal it receives. Wrong attribution = wrong bidding behavior

Attribution is not:

A way to inflate reported conversions (total conversions don't change, just how credit is distributed)
The same as a conversion window (window = how long after a click a conversion is counted; attribution = how credit is split)
Cross-channel attribution (Google Ads attribution only covers Google touchpoints — it doesn't natively see Meta, email, or organic)

归因是决定转化发生时哪些广告交互能获得功劳的规则。

它有两个核心作用：

决定报表展示内容 —— 哪些广告系列、关键词、广告看起来带来了效果
直接为Smart Bidding提供数据输入 —— 算法会基于接收到的信号优化，归因错误会直接导致出价行为错误

归因不是：

虚高上报转化数的手段（总转化数不会变化，变化的只是功劳分配方式）
等同于转化窗口（窗口定义点击后多长时间内的转化会被统计，归因定义功劳如何拆分）
等同于跨渠道归因（Google Ads归因仅覆盖Google触点，原生不识别Meta、邮件、自然流量触点）

Attribution Models

归因模型

Last Click (Google Ads default — legacy)

最后点击（Google Ads默认 —— 遗留模型）

100% of credit goes to the last ad click before conversion.

When it's appropriate:

Short purchase cycles where the last click genuinely drove the decision
Single-campaign accounts with no multi-touch complexity
When you're just getting started and conversion volume is low

The trap: Brand campaigns almost always get the last click. Last-click attribution makes brand campaigns look like your best performer — they're capturing intent created by other campaigns, not creating it. This causes under-investment in prospecting and display.

100%的转化功劳分配给转化前的最后一次广告点击。

适用场景：

购买周期短，最后一次点击确实是促成转化的核心决策因素
单广告系列账户，没有多触点复杂场景
刚起步，转化量较低的阶段

常见陷阱： 品牌广告系列几乎总是获得最后点击，最后点击归因会让品牌广告看起来表现最好，但实际上它们只是收割了其他广告系列创造的用户意向，而非自己创造了需求。这会导致你对新客拓展和展示广告的投入不足。

Data-Driven Attribution (DDA) — Recommended default

数据驱动归因（DDA）—— 推荐默认选项

Uses machine learning to assign fractional credit to every ad interaction based on how each touchpoint actually contributed to conversion probability.

How it works: Google compares conversion paths that converted vs. paths that didn't, and calculates the incremental contribution of each touchpoint. A click early in the path that increased conversion probability by 30% gets more credit than one that only increased it 5%.

Requirements:

Minimum 300 conversions in the last 30 days for the conversion action
Minimum 3,000 ad interactions in the last 30 days
If thresholds aren't met, Google falls back to last click for that conversion action

Why DDA is better for Smart Bidding: Smart Bidding uses attribution signals to set bids. DDA gives the algorithm a more accurate picture of which keywords and audiences contributed to conversion — leading to better bid decisions upstream in the funnel.

When DDA may mislead:

Accounts with very low conversion volume (below DDA thresholds)
When the model doesn't have enough data to be reliable — check "Model status" in Conversion Actions

使用机器学习，基于每个触点对转化概率的实际贡献，为所有广告交互分配分数化的功劳。

工作原理： Google对比发生转化和未发生转化的用户路径，计算每个触点的增量贡献。如果路径早期的某一次点击将转化概率提升了30%，它获得的功劳会比仅提升5%的点击更高。

使用要求：

对应转化操作过去30天内至少有300次转化
过去30天内至少有3000次广告交互
如果不满足阈值，Google会自动回退到最后点击归因

DDA对Smart Bidding更友好的原因： Smart Bidding使用归因信号设置出价，DDA能为算法提供更准确的关键词、受众转化贡献画像，帮助漏斗上游环节做出更好的出价决策。

DDA可能失真的场景：

转化量极低的账户（低于DDA阈值）
模型没有足够数据支撑可靠性时 —— 可在转化操作页面查看「模型状态」确认

Linear

线性

Splits credit equally across all clicks in the conversion path.

When useful: For comparing "what if we treated every touchpoint equally" — primarily useful as a diagnostic comparison, not as a production attribution model.

将转化功劳平均分配给转化路径中的所有点击。

适用场景： 用于对比「如果我们平等对待所有触点会怎么样」，主要作为诊断对比工具，不建议作为生产环境的归因模型。

Time Decay

时间衰减

More credit to touchpoints closer in time to the conversion.

When appropriate: Very short sales cycles (same-day decisions) where recency genuinely indicates contribution.

Limitation: Systematically undervalues awareness and upper-funnel campaigns that start the consideration process. Avoid for B2B with long sales cycles.

距离转化时间越近的触点，获得的功劳越多。

适用场景： 极短销售周期（当天决策），触点的时效性确实和贡献直接相关的场景。

局限性： 会系统性低估开启用户考虑流程的品牌曝光和上层漏斗广告，不适合销售周期长的B2B业务。

Position-Based (40/20/40)

位置基于（40/20/40）

40% credit to first click, 40% to last click, 20% split across middle touchpoints.

When appropriate: When you want to value both acquisition (first touch) and conversion (last touch) equally, and your account has clear prospecting and retargeting campaigns with a linear funnel.

40%功劳给首次点击，40%给最后点击，剩余20%平均分配给中间触点。

适用场景： 你希望同等重视获客（首次触点）和转化（末次触点），且账户有清晰的新客拓展和重定向广告系列，漏斗路径呈线性的场景。

First Click

首次点击

100% credit to the first ad interaction.

Rarely used in production. Useful as a diagnostic to see which campaigns initiate journeys — but systematically undervalues closing campaigns.

100%功劳分配给首次广告交互。

生产环境极少使用。 可作为诊断工具查看哪些广告系列开启了用户转化路径，但会系统性低估收口类广告系列的价值。

Choosing the Right Attribution Model

选择合适的归因模型

Scenario	Recommended Model
300+ conversions/mo, Smart Bidding active	Data-Driven Attribution
<300 conversions/mo	Last Click (DDA unreliable at low volume)
Long B2B sales cycle (14+ days)	Data-Driven or Position-Based
Pure brand campaign only	Last Click is fine — single touchpoint anyway
Diagnosing brand vs prospecting credit	Run model comparison before changing anything

The model comparison workflow: Before switching models, pull the "Attribution" report in Google Ads (Tools → Attribution) and run a model comparison. See how conversion credit shifts before committing — don't change attribution on live Smart Bidding campaigns without understanding the downstream bid impact.

场景	推荐模型
月转化300次以上，已开启Smart Bidding	数据驱动归因
月转化低于300次	最后点击（低数据量下DDA不可靠）
长周期B2B销售（14天以上）	数据驱动归因或位置基于
仅投放纯品牌广告系列	最后点击即可 —— 本身只有单一触点
诊断品牌和新客拓展广告的功劳分配	调整前先运行模型对比

模型对比工作流： 切换模型前，先拉取Google Ads中的「归因」报表（工具 → 归因）运行模型对比，确认转化功劳的偏移情况再落地调整。不要在没有评估下游出价影响的情况下，直接修改正在运行的Smart Bidding广告系列的归因设置。

Attribution Windows

归因窗口

Attribution windows control how long after a click (or view) a conversion is still credited to that ad.

归因窗口控制点击（或浏览）广告后多长时间内的转化，仍会被计入该广告的功劳。

Click-through conversion window

点击转化窗口

Default: 30 days. Can be set to 1, 7, 14, 30, or 60 days.

How to choose:

Short cycle (same-day e-com): 7 days is usually sufficient
Considered purchase (SaaS trial → paid): 30 days
Long B2B sales cycle: 60 days (maximum) — but understand this means slower data feedback

The tradeoff: Longer windows capture more conversions accurately but delay optimization data. If a conversion happens 45 days after a click and your window is 30 days, it's invisible to the algorithm.

默认值：30天，可设置为1、7、14、30或60天。

选择方法：

短转化周期（当天转化的电商）：7天通常足够
决策型购买（SaaS试用 → 付费）：30天
长周期B2B销售：60天（最大值） —— 但需要了解这会导致数据反馈速度变慢

权衡点： 更长的窗口能更准确地捕获更多转化，但会延迟优化数据反馈。如果点击后45天才发生转化，而你的窗口是30天，该转化对算法来说就是不可见的。

View-through conversion window

浏览转化窗口

Counts a conversion if a user saw (but didn't click) your Display or Video ad, then converted later via another channel.

Default: 1 day. Can be set to 1-30 days.

Important: View-through conversions are cross-device and require a leap of attribution faith — the user saw the ad, didn't click, and still converted. Use with caution:

Don't optimize Smart Bidding primarily on view-through conversions
Count them as informational signal, not primary conversion metric
For brand awareness measurement they're useful; for direct response bidding they can inflate reported performance

如果用户看过（但未点击）你的展示或视频广告，之后通过其他渠道转化，会被计入该广告的浏览转化。

默认值：1天，可设置为1-30天。

注意： 浏览转化是跨设备统计的，对归因的准确性要求更高 —— 用户看过广告、没点击、最终仍转化，使用时需谨慎：

不要主要基于浏览转化优化Smart Bidding
将其作为参考信息，而非核心转化指标
用于品牌曝光衡量时有效，但用于直接响应出价时可能虚高上报效果

Engaged view conversion window (Video)

互动视图转化窗口（视频）

For YouTube skippable in-stream ads: user watched 10+ seconds, didn't click, then converted. Default: 3 days.

针对YouTube可跳过贴片广告：用户观看了10秒以上、未点击，之后发生转化会被计入。默认值：3天。

How Attribution Directly Impacts Smart Bidding

归因对Smart Bidding的直接影响

This is the most important and least understood connection in Google Ads:

Smart Bidding trains on the conversion signal it receives. If your attribution gives 100% credit to last-click brand keywords, the algorithm learns to over-bid on brand keywords and under-bid on the non-brand keywords that actually created demand.

这是Google Ads中最重要也最容易被误解的关联逻辑：

Smart Bidding基于接收到的转化信号训练模型。 如果你的归因将100%功劳分配给最后点击的品牌关键词，算法就会学习到要对品牌关键词出高价，对真正创造需求的非品牌关键词出价不足。

Common misalignment patterns:

常见错位问题：

Pattern 1: Last-click + Smart Bidding overvalues brand

Brand campaign appears to have $12 CPA
Non-brand appears to have $58 CPA
Reality (via DDA): Both have similar contribution, brand is just closing journeys non-brand started
Effect: Algorithm over-invests in brand, under-invests in prospecting
Fix: Switch to DDA; bids will rebalance

Pattern 2: Short click window misses conversions

B2B SaaS with 21-day average sales cycle
Click window set to 7 days
Result: Algorithm thinks many clicks produced 0 conversions; bids down on keywords that actually convert
Fix: Extend window to 30-60 days; watch for conversion volume to increase in reports

Pattern 3: View-through conversions inflating CPA-target campaigns

Display campaigns optimizing to tCPA with view-through conversions included
CPA looks good but real CPA (click-through only) is 3× higher
Fix: Exclude view-through from primary bidding signal; measure separately

问题1：最后点击+Smart Bidding高估品牌价值

品牌广告系列看起来CPA是12美元
非品牌广告系列看起来CPA是58美元
实际情况（通过DDA查看）：两者贡献接近，品牌只是完成了非品牌开启的转化路径的收口
影响：算法对品牌投入过多，对新客拓展投入不足
修复方案：切换为DDA，出价会自动重新平衡

问题2：过短的点击窗口遗漏转化

B2B SaaS产品平均销售周期21天
点击窗口设置为7天
结果：算法认为很多点击没有带来转化，对实际能带来转化的关键词降低出价
修复方案：将窗口延长到30-60天，报表中的转化量会有所上升

问题3：浏览转化虚高CPA目标广告系列效果

展示广告系列优化tCPA时包含了浏览转化
报表上的CPA看起来很好，但实际仅统计点击的真实CPA是3倍以上
修复方案：将浏览转化从核心出价信号中排除，单独统计

Conversion Path Analysis

转化路径分析

The Attribution reports in Google Ads show you the actual paths users take.

Where to find: Tools → Attribution → Paths, Assisted Conversions, Model Comparison

Google Ads的归因报表可以展示用户的真实转化路径。

入口： 工具 → 归因 → 路径、辅助转化、模型对比

Assisted Conversions Report

辅助转化报表

Shows how many conversions each campaign/keyword "assisted" (appeared in the path but wasn't the last click).

Key metric: Assisted/Last-Click conversion ratio

Ratio > 1.0: Campaign assists more than it closes → typically upper-funnel campaign
Ratio < 1.0: Campaign closes more than it assists → typically lower-funnel, retargeting, or brand
Ratio ≈ 1.0: Campaign plays both roles equally

Action: Don't cut campaigns with high assist ratios just because their last-click ROAS looks poor. They may be feeding your closers.

展示每个广告系列/关键词「辅助」了多少次转化（出现在转化路径中但不是最后点击）。

核心指标： 辅助/最后点击转化比值

比值 > 1.0：广告系列的辅助作用多于收口作用 → 通常是上层漏斗广告系列
比值 < 1.0：广告系列的收口作用多于辅助作用 → 通常是下层漏斗、重定向或品牌广告系列
比值 ≈ 1.0：广告系列同等承担两种角色

行动建议： 不要仅因为高辅助比值的广告系列最后点击ROAS看起来差就停掉它们，它们可能正在为你的收口广告系列输送流量。

Top Paths Report

Time Lag Report

时间滞后报表

Shows how long after the first click conversions tend to happen.

Use for:

Validating your click window setting (if 20% of conversions happen after day 30, your 30-day window is losing them)
Setting client expectations on when to evaluate new campaign performance
Understanding how long Smart Bidding needs to learn before results stabilize

展示首次点击后通常过多久会发生转化。

用途：

验证你的点击窗口设置是否合理（如果20%的转化发生在30天之后，你的30天窗口就会遗漏这些转化）
给客户设定期望，明确评估新广告系列效果的合理时间
了解Smart Bidding需要多久的学习期才能让结果稳定

Google Analytics 4 vs Native Google Ads Conversion Tracking

Google Analytics 4 vs Google Ads原生转化跟踪

A critical attribution decision: which conversion source to use?

	Native Google Ads Tracking	GA4 Imported Goals
Coverage	Google Ads clicks only	All sessions (organic, direct, email, etc.)
Smart Bidding compatibility	Full	Full (when imported properly)
Cross-channel view	No	Yes
Attribution model	Google Ads models	GA4 data-driven (cross-channel)
Best for	Google Ads optimization	Full-funnel reporting

Recommendation: Use native Google Ads tracking as your primary Smart Bidding signal. Use GA4 imported conversions as a secondary signal or for reporting cross-channel truth.

Do not import GA4 goals as your only conversion signal and then use last-click attribution in GA4 — you'll feed the algorithm a distorted view of cross-channel performance.

一个关键的归因决策：使用哪种转化数据源？

	Google Ads原生跟踪	GA4导入目标
覆盖范围	仅覆盖Google Ads点击	覆盖所有会话（自然流量、直接访问、邮件等）
Smart Bidding兼容性	完全兼容	正确导入的情况下完全兼容
跨渠道视图	无	有
归因模型	Google Ads模型	GA4数据驱动（跨渠道）
适用场景	Google Ads优化	全漏斗报表

建议： 使用Google Ads原生跟踪作为核心Smart Bidding信号，使用GA4导入转化作为辅助信号或跨渠道真实效果报表数据源。

不要将GA4目标作为唯一转化信号，同时在GA4中使用最后点击归因 —— 这会给算法输入被扭曲的跨渠道表现数据。

Optimization Checklist

优化清单

When setting up or auditing

设置或审计时

Monthly

月度检查

Pull Assisted Conversions report — flag any "low-performing" campaigns that have high assist ratios
Review Time Lag report — is the click window capturing 90%+ of conversions?
Check for new conversion actions added without attribution settings reviewed

拉取辅助转化报表 —— 标记那些看起来「表现差」但辅助比值很高的广告系列
查看时间滞后报表 —— 当前点击窗口是否覆盖了90%以上的转化？
检查是否有新增的转化操作没有审核归因设置

Quarterly

季度检查

Re-run model comparison — does credit distribution still make sense?
Review if DDA thresholds are now met for conversion actions previously on last-click
Check GA4 vs Google Ads conversion totals for discrepancy investigation

重新运行模型对比 —— 当前的功劳分配逻辑是否仍然合理？
检查之前使用最后点击的转化操作是否已经满足DDA阈值
对比GA4和Google Ads的转化总数，排查数据差异

Common Mistakes

常见错误

Switching attribution models on live Smart Bidding campaigns without a transition plan Changing from last-click to DDA shifts conversion credit significantly. The algorithm re-learns, which can trigger a learning period and temporary performance dip. Best practice: test with a campaign experiment first, or switch during a low-stakes period.

Treating assisted conversions as "bonus" conversions Assisted conversions are not additional conversions — they represent the same conversions, viewed from different angles. Don't sum last-click + assisted; you'll double-count.

Setting a 30-day click window for a same-day purchase product If users typically buy within hours of clicking, a 30-day window is fine but doesn't capture more conversions — it just adds noise. Match window to actual behavior (Time Lag report tells you this).

Ignoring view-through conversion inflation Display campaigns that include view-through conversions can look remarkably efficient. Check what % of reported conversions are view-through before trusting Display ROAS figures.

Assuming Google Ads attribution shows the full customer journey Google Ads attribution is Google-Ads-click-centric. It cannot see organic search touches, email touches, or Meta ad touches. For cross-channel truth, use GA4 or a dedicated attribution tool.

没有过渡计划就直接给运行中的Smart Bidding广告系列切换归因模型 从最后点击切换到DDA会大幅改变转化功劳分配，算法需要重新学习，可能会触发学习期和临时效果下滑。最佳实践：先通过广告系列实验测试，或者在业务低峰期切换。

将辅助转化当做「额外」转化 辅助转化不是新增的转化 —— 它们是同一批转化从不同角度的呈现，不要把最后点击转化和辅助转化相加，会导致重复统计。

给当天就能完成购买的产品设置30天点击窗口 如果用户通常在点击后几小时内购买，30天窗口不会有坏处但也不会捕获更多转化，只会增加噪音。根据用户实际行为设置窗口（时间滞后报表可以提供数据支撑）。

忽略浏览转化的虚高问题 包含浏览转化的展示广告系列看起来效率会非常高，信任展示广告ROAS数据前，先查看浏览转化在总上报转化中的占比。

认为Google Ads归因能展示完整的用户旅程 Google Ads归因是以Google Ads点击为中心的，它看不到自然搜索、邮件、Meta广告的触点。要了解跨渠道真实情况，请使用GA4或专门的归因工具。