google-ads-attribution

name	google-ads-attribution
description	When the user wants help with Google Ads attribution models, data-driven attribution, attribution windows, cross-channel attribution, how attribution affects Smart Bidding, multi-touch conversion paths, or understanding which campaigns are actually driving conversions. Triggers on 'attribution', 'attribution model', 'data-driven attribution', 'last click attribution', 'assisted conversions', 'conversion window', 'multi-touch', 'attribution model comparison', 'first click attribution', 'credit allocation', or 'which campaign is driving conversions'. For conversion tracking setup see google-ads-conversion-tracking. For Smart Bidding strategy see google-ads-bidding.
metadata	{"version":"1.0.0"}

Google Ads — Attribution

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.

Before Starting

Check for product marketing context first: If .agents/product-marketing-context.md exists, read it before asking questions.

Gather this context:

1. Business and Funnel Context

• 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)

2. Current Attribution Setup

• 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)

3. Goals

• 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?

---

What Attribution Is — and Isn't

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

It does two things:

1. Determines what you see in reports — which campaigns, keywords, and ads look like they're driving results
2. 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)

---

Attribution Models

Last Click (Google Ads default — legacy)

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.

---

Data-Driven Attribution (DDA) — Recommended default

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

---

Linear

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.

---

Time Decay

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.

---

Position-Based (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.

---

First Click

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.

---

Choosing the Right Attribution Model

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.

---

Attribution Windows

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

Click-through conversion window

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.

View-through conversion window

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

Engaged view conversion window (Video)

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

---

How Attribution Directly Impacts Smart Bidding

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.

Common misalignment patterns:

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

---

Conversion Path Analysis

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

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

Assisted Conversions Report

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.

Top Paths Report

Shows the most common sequences of clicks before conversion.

What to look for:

• How many touchpoints on average? (1 = simple, linear; 4+ = complex, multi-channel)
• Which campaign types appear at the start of paths vs end?
• Does a specific combination of campaign types always appear in converting paths?

Time Lag Report

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

---

Google Analytics 4 vs Native Google Ads Conversion Tracking

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.

---

Optimization Checklist

When setting up or auditing

• Check attribution model per conversion action (Tools → Conversions → click conversion action → Settings)
• Verify click window matches typical sales cycle length
• Check DDA model status — is it "Active" or "Not enough data"?
• Run model comparison before switching any model on a live Smart Bidding campaign
• Confirm view-through conversions are not included in primary tCPA/tROAS bidding signal

Monthly

• 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

Quarterly

• 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

---

Common Mistakes

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.

---

Related Skills

• google-ads-conversion-tracking: Setting up conversion actions, tags, and tracking — the prerequisite for attribution to work correctly
• google-ads-bidding: Smart Bidding uses attribution signals directly — wrong attribution = wrong bids
• google-ads-audiences: Remarketing and RLSA audiences help close users from assisted campaigns — attribution explains why retargeting converts well
• google-ads-pmax: PMax has its own attribution behavior — conversions may be pulled from other campaign types depending on settings