Attribution is the science of deciding which marketing touchpoints deserve credit for a conversion. When a customer sees a Facebook ad on Monday, clicks a Google Ads search ad on Wednesday, and converts after clicking a remarketing banner on Friday — which channel gets the sale?
The answer depends entirely on which attribution model you use — and GA4 gives you several options. Getting this wrong means making budget decisions based on a distorted picture of which channels actually drive revenue.
This guide explains every GA4 attribution model, how data-driven attribution works in practice, and how to choose between them.
How Attribution Works in GA4
GA4 tracks the full path of touchpoints a user interacts with before converting. A “touchpoint” is any session where the user arrived via a trackable source — a Google Ads click, an email link, an organic search result, a social post.
GA4 stores up to 30 days of touchpoints before a conversion (configurable to 7 or 90 days in attribution settings). When a conversion happens, GA4 distributes credit across these touchpoints according to the selected model.
Where to Find Attribution Settings
GA4 Admin → Attribution Settings
Here you configure:
- Reporting attribution model — used across all GA4 reports
- Lookback window — how far back GA4 looks for touchpoints (7, 30, or 90 days for acquisition; 1 or 7 days for engagement and ad clicks)
Official reference: GA4 attribution models
The Attribution Models Available in GA4
Data-Driven Attribution (Default)
How it works: Machine learning analyses your actual conversion paths and assigns fractional credit to each touchpoint based on its incremental contribution to conversions. Google’s model uses counterfactual analysis — it compares paths that converted with similar paths that didn’t, and identifies which touchpoints made the difference.
Credit distribution example:
- User path: Organic Search → Email → Google Ads → Purchase
- Data-driven might assign: Organic 20%, Email 15%, Google Ads 65%
The exact distribution changes over time as the model learns from your data.
Requirements: Data-driven attribution requires a minimum volume of conversion events (typically 400+ conversions per month) to build a reliable model. Properties with lower conversion volume fall back to Last Click.
Best for: Most properties — it’s the most accurate model when you have sufficient conversion volume because it’s based on your actual data rather than a predetermined rule.
Caution: It’s a black box. You can’t see exactly why credit was distributed the way it was, which makes it hard to explain to clients or stakeholders who want a deterministic answer.
Last Click
How it works: 100% of conversion credit goes to the last touchpoint before the conversion.
Credit distribution example:
- User path: Organic Search → Email → Google Ads → Purchase
- Last Click assigns: Organic 0%, Email 0%, Google Ads 100%
Best for: Simple reporting where you want to credit the final decision-driving touchpoint. Often used for paid search analysis since search ads frequently appear at the bottom of the funnel.
Problem: It completely ignores all upper-funnel activity. Your content marketing, social media, and email nurture sequences get zero credit even if they were instrumental in creating awareness and intent.
First Click
How it works: 100% of conversion credit goes to the first touchpoint in the conversion path.
Credit distribution example:
- User path: Organic Search → Email → Google Ads → Purchase
- First Click assigns: Organic 100%, Email 0%, Google Ads 0%
Best for: Understanding which channels are best at acquiring new customers who eventually convert.
Problem: It ignores all the nurturing and retargeting that actually brought the customer to conversion. A brand running heavy retargeting will see those campaigns get no credit.
Linear
How it works: Equal credit distributed across all touchpoints.
Credit distribution example:
- User path: Organic Search → Email → Google Ads → Purchase (3 touchpoints)
- Linear assigns: Organic 33.3%, Email 33.3%, Google Ads 33.3%
Best for: Getting a broad view of channel contribution across the whole path. Useful when you believe every touchpoint contributes equally and you want to avoid over- or under-crediting any channel.
Problem: It’s not realistic — not every interaction contributes equally. The click that brought someone to a product page for the first time isn’t the same as the remarketing click that brought them back to complete the purchase.
Position Based (U-Shaped)
How it works: 40% credit to the first touchpoint, 40% to the last touchpoint, and the remaining 20% shared equally among middle touchpoints.
Credit distribution example:
- User path: Organic Search → Email → Google Ads → Purchase
- Position-based: Organic 40%, Email 10% (20% ÷ 2 middle steps, shared with… wait, Email is middle), Google Ads 40%
Actually with 3 touchpoints:
- First (Organic): 40%
- Middle (Email): 20%
- Last (Google Ads): 40%
Best for: Organisations that value both acquisition (first touch) and conversion (last touch) channels but also want some credit for nurturing channels in between.
Time Decay
How it works: More credit to touchpoints that occurred closer to the conversion. Credit decreases exponentially as touchpoints get further back in time (half-life of 7 days by default).
Credit distribution example (3-day path):
- Day 1: Organic Search → receives least credit
- Day 2: Email → receives moderate credit
- Day 3: Google Ads (day before conversion) → receives most credit
Best for: Short sales cycles where recency is a strong signal of influence. If your typical purchase cycle is 2–3 days, recent touchpoints genuinely are more influential.
Problem: For long purchase cycles (weeks or months), this model heavily penalises awareness channels even if they were instrumental in starting the journey.
How Data-Driven Attribution Actually Works
Understanding the mechanics helps you trust (or question) the model’s output.
The Shapley Value Approach
GA4’s data-driven model uses a concept from cooperative game theory called Shapley values. The idea:
Imagine all your marketing channels as “players” in a cooperative game where the “prize” is conversions. The Shapley value calculates each player’s contribution by looking at every possible ordering of the players and measuring how much each one adds to the group’s performance.
In practice, GA4:
- Identifies all conversion paths in your data
- Groups paths by their touchpoint combinations
- Compares conversion rates of paths with a given touchpoint vs. similar paths without it
- Assigns credit proportional to each touchpoint’s incremental conversion contribution
What This Means for Your Reporting
Upper-funnel channels get more credit than with Last Click — blog content, organic social, and display awareness campaigns that appear early in paths will show higher conversion contribution in data-driven attribution than in Last Click.
Branded search may get less credit — if branded search always appears last in the path (users Google your brand name to come back and buy), data-driven may assign less credit to branded search than Last Click does, because the model recognises the conversion would have happened anyway based on upper-funnel intent.
The model needs time — data-driven attribution looks back at the prior 30 days. If you’ve just changed your channel mix significantly, the model’s credit assignments may lag reality for a few weeks while it recalibrates.
GA4 Attribution vs. Google Ads Attribution
An important distinction: GA4 attribution (what you see in GA4 reports) and Google Ads attribution (what Google Ads uses for bidding and optimisation) are separate systems.
| GA4 Attribution | Google Ads Attribution | |
|---|---|---|
| Scope | All channels (organic, email, social, paid) | Google Ads channels only |
| Used for | GA4 conversion reports, channel reporting | Smart bidding, campaign optimisation |
| Where to configure | GA4 Admin → Attribution Settings | Google Ads → Conversion Actions |
| Imports data from | GA4 → linked to Google Ads | Separate from GA4 attribution |
When GA4 conversions are imported into Google Ads, Google Ads applies its own attribution model for bidding purposes. Changing your GA4 attribution model doesn’t change how Google Ads Smart Bidding works.
Cross-Channel Attribution: The Limitation
GA4 only attributes across channels it can see. It cannot attribute across:
- Offline channels — TV, radio, print, events
- Channels without UTM tracking — dark social (WhatsApp, Slack), many email clients
- Channels that bypass GA4 — users with ad blockers, incognito sessions
For businesses with significant offline spend or dark social traffic, GA4’s attribution will systematically over-credit trackable digital channels at the expense of untrackable ones.
Which Attribution Model Should You Use?
Use Data-Driven if: You have 400+ monthly conversions and want the most accurate model. This should be your default.
Use Last Click if: You’re specifically analysing paid search performance or reporting to a client who uses Last Click as their standard. Also useful as a comparison point to see how data-driven differs.
Use First Click if: Your primary question is “what channel first introduces customers to us?” — useful for content marketing or brand awareness reporting.
Use Linear if: You’re doing a high-level channel audit and want every touchpoint visible without over-crediting any single channel.
Use Time Decay if: Your purchase cycle is under a week and you want to credit the channels that pushed users to the final decision.
Comparing Models Side by Side in GA4
GA4 doesn’t have a built-in model comparison tool (unlike Universal Analytics). The workaround:
- Advertising → Attribution → Model Comparison report (if you have Google Ads linked)
Or, if you’re using BigQuery:
-- Compare first touch vs last touch attribution using raw events
WITH touchpoints AS (
SELECT
user_pseudo_id,
event_timestamp,
traffic_source.source AS source,
traffic_source.medium AS medium,
ROW_NUMBER() OVER (
PARTITION BY user_pseudo_id
ORDER BY event_timestamp ASC
) AS touch_order,
COUNT(*) OVER (PARTITION BY user_pseudo_id) AS total_touches
FROM `project.analytics_XXXXXXXXXX.events_*`
WHERE _TABLE_SUFFIX >= FORMAT_DATE('%Y%m%d', DATE_SUB(CURRENT_DATE(), INTERVAL 30 DAY))
AND traffic_source.source IS NOT NULL
)
SELECT
source,
medium,
-- First touch: count when this is touch #1
COUNTIF(touch_order = 1) AS first_touch_conversions,
-- Last touch: count when this is the final touch
COUNTIF(touch_order = total_touches) AS last_touch_conversions
FROM touchpoints
GROUP BY 1, 2
ORDER BY last_touch_conversions DESC;
This lets you see exactly how first-touch vs last-touch credit differs by channel.
Understanding attribution is fundamental to making good media spend decisions. If your current reporting doesn’t reflect the full conversion path, you’re likely over-investing in bottom-funnel channels and under-investing in the awareness and consideration content that builds demand.
Our analytics team audits attribution setups and helps clients understand their real channel contribution using GA4 and BigQuery. Book a free consultation to discuss your attribution setup.