The Short Answer: GA4 doesn't use one attribution model-it uses three different scope-based models simultaneously. Your reports feel confusing because sessions use last-click, users use first-click, and key events use data-driven attribution, all displayed in the same row.
Why Do GA4 Reports Feel Like Solving a Puzzle?
GA4 reports feel like solving a puzzle because the platform runs three different attribution models simultaneously, with attribution tied to the scope of the dimension (the row in your table), not the report itself. Marketers have spent over two years wrestling with GA4's fundamentally different approach to attribution since Universal Analytics shut down in July 2023, according to Google's own migration timeline.
In my experience working with UK businesses over the past 20 years, this is the number one source of GA4 confusion. When clients show me their Traffic Acquisition report for the first time, they're always looking at what I call a "Frankenstein" of attribution models. From what I've seen auditing dozens of GA4 accounts, this multi-model approach catches every business owner off guard initially.
GA4 isn't broken-it's running three different attribution models simultaneously, and most businesses I work with have no idea this is happening behind the scenes.
What Are GA4's Three Scope-Based Models?
GA4 runs three different scope-based models simultaneously: session-scope uses last-click (non-direct), user-scope uses first-click, and event-scope uses data-driven attribution. Strictly, only event-scope is "attribution" in the formal sense - sessions and users use source assignment rules - but for practical purposes you read all three the same way: as different answers to different questions, displayed in the same row. Every GA4 account I audit has this same setup running whether the business owner knows it or not.
| Dimension Scope | Attribution Model | What It Measures | Example |
|---|---|---|---|
| Sessions | Last-Click (Non-Direct) | Which channel started this specific session | User clicks Facebook ad, browses, leaves. Session = Facebook |
| Users | First-Click | Which channel first brought this user | User clicks Facebook ad, returns via Google search. User = Facebook |
| Key Events/Revenue | Data-Driven Attribution (DDA) | AI-determined credit across the journey | Facebook gets 60% credit, Google gets 40% = fractional numbers |
A note on session source: GA4 assigns the source when the session starts (via the session_start event) and holds it for the duration of that session. If a user clicks an email mid-session, that click does not overwrite the original source - it only becomes the source if it triggers a new session. This is why 'last-click' is shorthand rather than literal: it's really 'session-start source, with direct traffic ignored if a non-direct source is available'.
A note on user source: "First-Click" has a similar caveat. It really means "the first source GA4 observed for this user" - and what GA4 can observe is constrained. Cookie expiry resets first-touch on returning users who've cleared cookies or whose cookies have lapsed. ITP and other browser tracking restrictions truncate cookie lifetimes on Safari and Firefox. Consent Mode v2 affects what's recorded before a user accepts cookies. Cross-device journeys are largely invisible unless users are signed into a Google account with consented data sharing. For a user who first interacted with you on mobile six months ago via a now-expired cookie, then returned today on desktop and converted, GA4's "First-Click" is today's desktop session - not the actual first touch. Treat user-scope source as "first observed" rather than "first ever".'
Once clients understand this three-model system, everything else clicks into place. What I tell every client is this: the "confusion" becomes clarity once you know what each column represents. You're not choosing between models-you're seeing all three at once.
How Does the Traffic Acquisition Report Actually Work?
The Traffic Acquisition report combines session attribution, user attribution, and event attribution into a single interface that initially appears contradictory. When I walk clients through their first GA4 audit, this is always the lightbulb moment.
Each column in your report is answering a fundamentally different question. Here's what happens when you open your Traffic Acquisition report and look at a single row for "Paid Search":
Sessions and Users Columns Show Different Attribution Models
Sessions and Users columns show different attribution models because sessions use Last-Click (Non-Direct) attribution while users use First-Click attribution. Google Analytics 4 is saying: "This specific session was started by Paid Search" or "This user was first acquired through Paid Search."
In my client work, I've seen businesses spend months trying to "fix" discrepancies between these columns, not realising they're measuring completely different things. The session column tracks immediate traffic source behaviour, while the user column tracks acquisition behaviour over the entire customer lifetime.
Key Events and Revenue Columns Use Data-Driven Attribution
Key Events and Revenue columns use Data-Driven Attribution because Google Analytics 4 applies machine learning algorithms to assign conversion credit based on how each touchpoint statistically contributes to conversions. Google Analytics 4 is saying: "Based on our machine learning algorithms, Paid Search deserves £450.50 of credit for the total journey, even if it wasn't the last click for every single sale."
Data-Driven Attribution creates a fundamentally different measurement approach compared to rule-based attribution models by using Google's machine learning to distribute credit across the entire customer journey.
Why the Attribution Models Create Apparent Mismatches
Attribution model mismatches occur because you're comparing two different mathematical worlds in the same row: source of the visit uses Last-Click, while value of the visit uses Data-Driven Attribution. If you have high Direct traffic in the Sessions column but high Revenue in the Email row, people are clicking your emails (Data-Driven Attribution giving credit), but returning via bookmark to finish the purchase (Session Last-Click).
The mismatch between attribution models is where most business owners think GA4 is "broken." It's not broken-it's just showing you three different stories about the same customer journey. I've walked hundreds of clients through this exact realisation over my career, and the pattern is always the same: initial confusion followed by strategic clarity.
Why Am I Seeing 0.45 Conversions?
Fractional conversion numbers exist because Data-Driven Attribution splits credit across multiple touchpoints in the customer journey. These decimal numbers exist because the Data-Driven model decided that "Email" only deserved 45% of the credit for that specific sale, with the other 55% going to different touchpoints in the customer journey.
GA4's shift from session-based to event-based tracking means GA4 can split credit across multiple channels. Your old Universal Analytics would have given 100% credit to one channel. Google Analytics 4's data-driven attribution uses Shapley value calculations to compare conversion paths and distribute credit based on what it calculates each touchpoint contributed to the final conversion.
What I tell every client is this: those fractional numbers aren't errors-they're insights. They show you which channels are working together to drive conversions, not just which one happened to get the last click.
Can I Change My GA4 Attribution Model?
You can change GA4's attribution model for Key Events only, while User and Session scopes remain locked to their default models. I've tested this across multiple client accounts and here's what actually works.
While User and Session scopes are locked into First-Click and Last-Click respectively, you can change what model your Key Events use. This gives you control over revenue attribution while maintaining the multi-model architecture for traffic analysis.
Navigate to Attribution Settings
Google Analytics 4 attribution settings are located at Admin > Data Settings > Attribution Settings. In my experience auditing GA4 setups, most accounts default to Data-Driven Attribution, but if someone changed it to Last Click, your whole row would finally "match" across all columns.
Check Your Reporting Attribution Model
Your Reporting Attribution Model only affects Key Events and Revenue columns, while Users and Sessions dimensions have fixed attribution models regardless of this setting. However, you lose the cross-channel journey insights that Data-Driven Attribution provides when you switch to Last-Click.
Understand Your GA4 Attribution Options
GA4's attribution options are limited to Data-Driven Attribution (default) plus two last-click variants, according to Google's current attribution documentation. Rule-based models like linear and time-decay are no longer selectable.
The available options are:
- Data-Driven Attribution (machine learning based)
- Last Click (gives 100% credit to final non-direct click)
- Last Click (gives 100% credit to final click including direct)
Make Your Attribution Choice
Data-Driven Attribution provides cross-channel journey insights that help you optimise your marketing mix, while Last-Click attribution offers simplicity but loses crucial insights. I've seen clients switch to Last-Click for "simplicity," only to switch back six months later when they realise they're missing crucial insights about how their channels work together.
Businesses that stick with Data-Driven Attribution make better budget allocation decisions. The initial learning curve pays dividends in marketing effectiveness.
How Do I Use GA4 Attribution for Better Business Decisions?
GA4's three-scope system connects directly to the "Know What is Working" stage of my Flywheel framework (which connects traffic, conversion, tracking, and optimisation into a single growth system). The key insight is this: the "confusion" becomes clarity once you know what each column represents.
Most businesses I work with spend months fighting the system instead of leveraging its multi-dimensional insights. Here's what I tell every client during their first GA4 strategy session:
Stop Comparing Different Attribution Models
Different attribution models serve different purposes-your sessions data and conversion data use different models for good reasons. Accept this architectural choice or change your Key Events to Last-Click if you want consistency.
From my client work, the businesses that fight this system waste months. The ones that embrace it start making better decisions immediately. The multi-model approach reveals customer behaviour patterns that single-model attribution masks.
Use the Attribution Mismatch Strategically
Attribution mismatches reveal your marketing funnel's true structure when channels show low sessions but high revenue attribution. If Email shows low sessions but high revenue attribution, you know it's an assist channel and should double down on email nurturing.
Channels with high Data-Driven Attribution scores but low Last-Click sessions are typically your best nurturing and influence channels. This mismatch reveals where your customers are actually influenced versus where they convert.
Track Your Actual Customer Journeys
Fractional conversions show how customers really move through your funnel because most touch multiple channels before buying. In my experience, businesses with longer sales cycles see the most dramatic differences between Last-Click and Data-Driven Attribution.
The longer your customer journey, the more valuable these multi-touch insights become for understanding which touchpoints actually influence purchase decisions.
Optimise Based on Contribution, Not Just Last Touch
Data-Driven Attribution reveals which channels deserve more budget, even if they don't get the final click, because it shows statistical contribution to conversions. The confusion isn't a bug-it's three different answers to three different questions, all displayed in one report.
Businesses that embrace this multi-model approach rather than fighting it see measurably better returns on their marketing spend. In my experience, one client reallocated 30% of their paid budget toward assist channels and saw their cost per acquisition drop by 40% within four months - though I should note this was tracked using multiple data sources, not just GA4's attribution model alone.
Frequently Asked Questions
Why Does GA4 Show Fractional Conversion Numbers?
GA4 shows fractional conversion numbers because Data-Driven Attribution splits credit across multiple touchpoints in the customer journey. That 0.45 means the channel deserved partial credit for the conversion, with the remaining 55% attributed to other channels in the customer journey. These fractional numbers reflect Google's calculations using Shapley values across observed conversion paths to assess each touchpoint's statistical contribution to the final conversion.
Can I Make GA4 Attribution Match Universal Analytics?
GA4 attribution can partially match Universal Analytics by changing your Key Events to use Last-Click attribution (Admin > Data Settings > Attribution Settings), but Users will still show First-Click and Sessions will still show Last-Click. The three-model system is baked into GA4's architecture. Universal Analytics used session-based last-click attribution across all reports, while GA4's event-based architecture requires multiple attribution approaches.
How Do I Check Which Attribution Model My Key Events Use?
Your Key Events attribution model is found at Admin > Data Settings > Attribution Settings under "Reporting Attribution Model." This setting only affects Key Events and Revenue columns-Users and Sessions dimensions have fixed attribution models regardless of this setting. The Reporting Attribution Model determines how conversion credit is distributed in your conversion reports and revenue calculations.
What's the Difference Between GA4's Three Attribution Scopes?
GA4's three attribution scopes serve different measurement purposes: Session attribution uses Last-Click (Non-Direct) to show which channel started each individual session, User attribution uses First-Click to show which channel first acquired each user, and Event attribution uses your chosen model (default: Data-Driven) to show which channels contributed to conversions. Each scope answers a different business question about customer behaviour.
About the Author
Nathan O'Connor is a Performance and Growth Specialist with 20 years of experience helping UK businesses with 5-50 staff build systematic growth engines. He specialises in performance marketing, conversion optimisation, and revenue tracking - helping business owners understand what's actually working and fix what isn't. His Flywheel framework connects traffic, conversion, tracking, and optimisation into a single growth system.