A customer visits the site through a Google search, returns two weeks later via an email link, clicks a retargeting ad the following week and finally converts by typing the URL directly. Which channel gets credit for the sale? The answer depends entirely on the attribution model, and different models give radically different answers. The wrong model doesn’t just produce inaccurate reports. It leads to budget decisions that move money away from channels that work and toward channels that don’t.
At Gorilla Marketing, attribution is a practical concern because we manage both SEO and PPC for clients. When organic search introduces a customer and paid search closes them, the attribution model determines how each channel is valued. Getting it wrong means misallocating budget between the two. This guide covers how attribution models work, what GA4 actually supports now, and how to build a measurement approach that accounts for attribution’s real limitations.
What Attribution Modelling Actually Does
Attribution modelling assigns credit for conversions to the marketing touchpoints that preceded them. A touchpoint is any interaction a customer has with the business before converting: clicking an ad, opening an email, visiting from organic search, seeing a display ad or arriving through social media.
The model determines the rules for distributing credit. Some models give all credit to one touchpoint. Others spread it across the journey. The choice shapes how every channel appears in reports, which directly influences where marketing budget goes.
The average customer journey now involves approximately 6.5 touchpoints before conversion, with B2B journeys averaging 14 or more. With marketers widely reporting that customer paths are getting longer, the model you choose has more impact on your budget decisions than ever.
Single-Touch Models
Single-touch models assign 100% of credit to one touchpoint. Simple to understand but inherently incomplete because they ignore the rest of the journey.
Last click
All credit goes to the final touchpoint before conversion. If a customer found the site through organic search, returned via email and converted after clicking a Google Ad, the ad gets 100% credit.
When it works: Short sales cycles where the last interaction genuinely represents the decision. Quick e-commerce purchases, low-consideration products.
Where it fails: Any multi-channel journey. Last click overvalues closing channels (paid search, direct, remarketing) and undervalues channels that introduce customers (organic search, social, content marketing). For businesses running both SEO and PPC, last click makes SEO look less valuable than it is, because organic often starts journeys that paid closes. Display ads frequently appear in conversion paths as an assist but rarely as the final click, meaning last click massively understates their contribution. Despite these limitations, 22% of marketers still rely exclusively on last click. It’s the default for a reason (simplicity), but that simplicity comes at a cost: research shows 15 to 30% higher marketing ROI for businesses that move to more sophisticated attribution.
First click
All credit goes to the first touchpoint. If organic search was the first interaction, it gets 100% regardless of what happened afterwards.
When it works: Understanding acquisition. Which channels bring new customers into the funnel?
Where it fails: Ignores everything after discovery. Overvalues awareness channels, undervalues the nurturing and closing that actually converts prospects.
Multi-Touch Models
Multi-touch models distribute credit across multiple touchpoints. More realistic but harder to interpret. Worth noting upfront: GA4 removed most of these as reporting options in November 2023 (covered in the GA4 section below), but they remain important conceptual frameworks used in other platforms and custom attribution setups.
Linear
Equal credit to every touchpoint. Five touchpoints, 20% each.
Strengths: Acknowledges the full journey. Nothing gets ignored.
Weaknesses: Treats every interaction as equally important. A blog visit three months ago gets the same credit as the ad click that closed the sale yesterday. Produces reports where everything looks moderately useful and nothing stands out, which makes budget decisions harder rather than easier.
Time decay
More credit to touchpoints closer to conversion, less to earlier ones.
Strengths: Reflects the reasonable assumption that recent interactions matter more. Suits longer consideration periods where late-stage nurturing genuinely accelerates conversion.
Weaknesses: Undervalues top-of-funnel activity. If the initial discovery determines whether a customer enters the funnel at all, time decay understates the importance of awareness channels. SEO is typically the largest source of first-touch awareness, meaning time decay systematically undervalues organic search.
Position-based (U-shaped)
40% credit to the first touchpoint, 40% to the last, 20% distributed across everything in between.
Strengths: Values both introduction and closing while acknowledging the middle. A pragmatic compromise.
Weaknesses: The 40/40/20 split is arbitrary. No evidence that real customer journeys assign value this way. Businesses use it because the split feels right, not because data supports it.
Data-driven attribution
Machine learning analyses actual conversion paths to assign fractional credit based on which touchpoints statistically contribute to conversions. This is now GA4’s default and, in practice, the only multi-touch option GA4 offers.
Strengths: The only model that learns from the business’s own data rather than applying a fixed rule. Adapts to specific customer journeys. Companies using data-driven attribution report 1.7 times faster revenue growth than those using rules-based models.
Weaknesses: Requires volume. GA4 needs approximately 400 conversions per key event to run data-driven attribution reliably. Below that, it silently falls back to last click without notification. The methodology is opaque: Google doesn’t publish how credit is distributed, making it difficult to validate.
What GA4 Actually Supports Now
This is where many guides are outdated. In November 2023, Google removed first-click, linear, time-decay and position-based models from GA4. The only options remaining are data-driven attribution (the default) and last click.
Navigate to Admin, then Data Display, then Events, then Attribution Settings. The key controls:
Reporting attribution model. Data-driven or last click. That’s it. If your business previously relied on position-based or time-decay in GA4, you’ll need to either use data-driven or move to external attribution tools.
Lookback windows. The maximum time between a touchpoint and conversion for credit to be assigned. Default is 30 days for acquisition events and 90 days for other events. For B2B businesses with long sales cycles (the average is 92 days), extending this window is important. The default captures less than half of a typical B2B journey.
Channel exclusions. Remove specific channels from attribution. The most common use: excluding direct traffic, since direct visits often represent users who already decided to convert.
One critical detail most guides miss: GA4’s attribution settings only apply to dimensions that don’t have a “session” or “first user” prefix. If you report on “Session source/medium”, you’ll get last-touch attribution regardless of your model setting. Only “Source/medium” (without the session prefix) uses the configured model.
How Attribution Differs by Business Type
The right attribution approach depends not just on model choice but on the type of business and its customer journey characteristics.
B2B and professional services. Longer sales cycles (92 days on average), multiple stakeholders involved in buying decisions, and heavy reliance on content marketing for nurturing. B2B buyers consume an average of four or more content pieces before contacting sales. Data-driven attribution is the right model, but the lookback window must be extended well beyond GA4’s default. LinkedIn is often the dominant B2B first-touch channel, meaning last click will consistently undervalue its contribution. CRM integration is essential, because many B2B touchpoints happen off-site (email, calls, meetings) and need to be stitched into the attribution picture manually.
E-commerce and DTC. Shorter cycles but more complex channel mixes, particularly for businesses running paid social alongside search. 55% of paid social conversions require three or more touchpoints. Server-side tracking becomes critical here because cookie restrictions affect the purchase events that matter most. Post-purchase surveys (“How did you hear about us?”) are increasingly used as an attribution signal to validate what digital models show.
Local services. High proportion of offline conversions (phone calls, walk-ins) that digital attribution misses entirely. Call tracking with dynamic number insertion helps bridge the gap, but attribution for local businesses will always be partial without supplementing digital data with offline measurement.
How to Choose the Right Approach
Short sales cycle, low consideration: Last click works well enough. The closing interaction usually represents the decision.
Long cycle, high consideration: Data-driven, if you meet the volume threshold. If not, last click supplemented with manual review of conversion paths in GA4’s Advertising section.
Businesses with strong organic + paid closing: Data-driven is the best GA4 option. If SEO introduces customers and PPC closes them, last click will always overstate PPC’s contribution and understate organic’s role.
Insufficient GA4 volume: Consider external attribution platforms (Ruler Analytics, Dreamdata, HockeyStack for B2B) that bring their own models and can work with lower volumes.
Common Attribution Mistakes
Optimising for the model instead of the business. Switching from last click to data-driven and then reallocating budget purely based on the new numbers. The model changed, the business didn’t. Use model comparison for directional insights, not autopilot budget shifts.
Treating assisted conversions as less valuable. A channel appearing in 40% of conversion paths but rarely as the last touch is doing critical work. Cutting it because it doesn’t “close” is like removing the salesperson who books all the meetings because they don’t sign the contracts.
Ignoring the “not set” problem. GA4 frequently shows “(not set)” or “Unassigned” in attribution reports, particularly with cross-domain tracking, consent gaps and misconfigured UTM parameters. If 15% of conversions show unattributed, the model is working with incomplete data. Fix data quality before trusting attribution output.
Never validating with real-world data. Attribution says organic drives 30% of revenue. Does that match what the sales team hears? Do customers mention Google when they call? Post-purchase surveys, even simple ones, provide a reality check that digital attribution can’t.
Setting the wrong lookback window. GA4’s default 30-day window works for impulse purchases. It doesn’t work for B2B, high-ticket e-commerce or professional services where the average time-to-close exceeds 60 days.
Where Attribution Breaks Down
Attribution models measure the digital, click-based journey. They have structural limitations that no model selection can fix.
Offline touchpoints are invisible. Phone calls, in-store visits, word-of-mouth referrals and trade show interactions don’t appear unless specifically tracked. For businesses with significant offline components, attribution tells only part of the story.
Impression influence is excluded. Attribution counts clicks, not views. A customer who saw a display ad five times but never clicked gives zero credit to display. The ads may have influenced the conversion, but click-based attribution can’t measure it.
Cross-device journeys break. Research on mobile, conversion on desktop. Often appears as two separate users. Cross-device attribution needs authenticated users, and 36% of consumers switch devices mid-journey.
Privacy and consent gaps. Users who decline tracking cookies are invisible to attribution entirely. iOS 14+ reduced observable conversions by 18 to 32%. As consent rates vary, the attributed journey is only a partial picture of the actual customer journey. Server-side tracking recovers some of this lost signal, but doesn’t eliminate the gap.
The Measurement Triangle: What Sits Alongside Attribution
Attribution on its own is increasingly insufficient. The strongest measurement frameworks combine three approaches, sometimes called the measurement triangle.
Multi-touch attribution (MTA) handles tactical, day-to-day channel analysis. It’s what we’ve covered above. Good for answering “which channels are contributing to conversions?” with granular, user-level data.
Marketing mix modelling (MMM) uses aggregate, time-series data (weekly spend, revenue, impressions, external factors like seasonality) to model channel contribution without any user-level tracking. It’s privacy-proof because it doesn’t rely on cookies or consent. Google released Meridian, an open-source MMM framework, in January 2025. Meta’s Robyn is another widely used option. MMM is best for strategic, quarterly budget allocation across channels, including offline channels that attribution can’t see.
Incrementality testing answers the causal question: “What would have happened if we hadn’t spent on this channel?” Using controlled experiments (geographic holdout tests are the most common), it measures the true incremental impact of a channel rather than relying on correlation. It’s the most rigorous approach but also the slowest and most expensive to run.
No single approach gives the full picture. Attribution provides granular, real-time channel data but misses offline and impression influence. MMM captures everything including offline but operates at a strategic level with limited granularity. Incrementality testing proves causation but only for one channel at a time. Used together, they calibrate and validate each other.
| Approach | Best For | Granularity | Privacy Impact | Cost |
|---|---|---|---|---|
| Multi-touch attribution | Tactical daily optimisation | User-level | High (cookie-dependent) | Low |
| Marketing mix modelling | Strategic quarterly allocation | Channel-level | None (aggregate data) | Medium |
| Incrementality testing | Causal validation | Campaign-level | Low | High |
For most SMEs, data-driven attribution in GA4 supplemented with periodic incrementality tests is a realistic starting point. Enterprise businesses with larger budgets and more complex channel mixes benefit from adding MMM into the framework.
Making Attribution Actionable
Attribution data should inform budget decisions, not make them automatically.
Compare how data-driven and last click value the same channels. If organic search gets 5% of credit under last click but 25% under data-driven, that gap reveals how much the model shapes perception. The truth is somewhere in the range, and the range itself is informative.
Look at conversion paths in GA4 (Advertising, then Conversion Paths) to see actual multi-touch journeys. Raw path data often tells a clearer story than any model’s summary. Which channel combinations appear most frequently before high-value conversions?
Run the analysis quarterly. Customer journeys change. Seasonal patterns shift attribution. A model that works in Q1 may misrepresent Q3 behaviour.
Don’t treat attribution as a scorecard for individual channels. Treat it as a tool for understanding how channels work together. The most common attribution mistake isn’t choosing the wrong model. It’s cutting budget from a channel that looks weak on last click without realising it’s doing essential top-of-funnel work that other channels depend on.
For businesses that want to go deeper, export GA4 data to BigQuery for proper session stitching and custom analysis. GA4’s interface provides a useful overview, but the raw data in BigQuery allows custom attribution windows, exclusion rules and channel groupings that the standard reports can’t support.
Gorilla Marketing’s analytics and tracking and digital strategy work includes attribution configuration, multi-channel analysis and budget recommendations based on data rather than default model assumptions. Get in touch to discuss how attribution is shaping your current marketing decisions.
