This guide covers the 5 main attribution models: last click, first click, linear, time decay, data-driven, plus which model to use for ecommerce and how to set it up in GA4 and Google Ads.
1. Last Click Attribution
Last click gives 100% of the conversion credit to the last touchpoint before the purchase. If a customer clicked a Google Shopping ad and bought, Google Shopping gets all the credit. Even if the customer first discovered your brand through a Facebook ad two weeks earlier.
This is the simplest model and the one most people default to. Google Ads and Meta both use variations of last click in their own reporting (though Meta uses a 7-day click, 1-day view window, which is slightly different).
The problem with last click for ecommerce: it overvalues bottom-funnel campaigns (branded search, retargeting) and undervalues top-funnel campaigns (prospecting, awareness). If you only look at last click, you might conclude that your Facebook prospecting campaigns are worthless, when actually they are feeding customers into the funnel that Google branded search later converts.
That said, last click is useful for one thing: it is the most conservative attribution model. If a campaign looks profitable on last click, it is almost certainly profitable. It is a good reality check when other models are painting a rosier picture.
2. First Click Attribution
First click is the opposite. It gives 100% credit to the first touchpoint. The channel that introduced the customer to your brand gets all the credit, regardless of what happened after.
This model overvalues awareness channels and undervalues conversion channels. It is rarely used as a primary model, but it is useful as a comparison. If your Facebook prospecting campaigns look terrible on last click but great on first click, that tells you something: they are good at introducing new customers who then convert through other channels.
For ecommerce, first click is most useful when you are trying to understand which channels drive new customer acquisition specifically. If you are measuring customer acquisition cost (CAC) rather than ROAS, first click gives you a more accurate picture of where new customers are coming from.
3. Linear Attribution
Linear attribution splits credit equally across all touchpoints. If a customer interacted with four channels before buying, each gets 25% credit.
The appeal of linear attribution is its fairness. Every touchpoint gets recognized. The downside is that it treats a random impression the same as the final click that drove the purchase. Not all touchpoints are equally valuable, and linear pretends they are.
For ecommerce stores running 3-5 channels, linear attribution can be useful for getting a general sense of channel contribution. It is better than last click at recognizing upper-funnel channels, and better than first click at recognizing lower-funnel ones. But it is a compromise that does not really reflect reality.
4. Time Decay Attribution
Time decay gives more credit to touchpoints closer to the conversion and less to earlier ones. A Facebook ad the customer clicked 14 days ago gets less credit than the Google Shopping ad they clicked today.
This is probably the most intuitive model for ecommerce. Recent interactions are usually more relevant to the buying decision than ones from weeks ago. A customer who clicked your retargeting ad yesterday and bought today was probably influenced more by that ad than by the blog post they read two weeks earlier.
The downside: it still undervalues top-of-funnel activity. Prospecting campaigns that introduce customers early in the journey get minimal credit because they are far from the conversion. If you use time decay exclusively, you might cut prospecting budgets that are actually driving growth.
5. Data-Driven Attribution
Data-driven attribution (DDA) uses machine learning to analyze your actual conversion paths and assign credit based on what is statistically likely to have influenced the conversion. It is available in both GA4 and Google Ads.
Unlike the rule-based models above, DDA adapts to your specific data. If it finds that customers who click a Facebook ad AND a Google Shopping ad convert at a much higher rate than those who only click one, it gives meaningful credit to both touchpoints. It learns patterns from your actual customer journeys.
The catch: DDA needs data to work. Google Ads requires at least 300 conversions in 30 days for reliable data-driven attribution. GA4 has its own thresholds. If you do not have enough data, DDA falls back to a model similar to linear attribution.
For ecommerce stores with 300+ monthly conversions, DDA is generally the best option. It is not perfect (no model is), but it adapts to your business instead of applying arbitrary rules.
6. Which Model Should Ecommerce Use?
Here is our general recommendation based on what we see work for ecommerce stores:
- If you have 300+ conversions/month: Use data-driven attribution in both GA4 and Google Ads. It is the most accurate for your specific business.
- If you have 50-300 conversions/month: Use time decay in GA4 and last click in Google Ads (for bidding). Time decay gives a more nuanced view for reporting, while last click is more conservative and stable for bid optimization with limited data.
- If you have under 50 conversions/month: Use last click everywhere. You do not have enough data for models that require statistical significance. Focus on getting more conversions first.
Regardless of which model you use for primary reporting, we suggest running a monthly comparison across models. Look at your top campaigns in last click, first click, and data-driven side by side. If a campaign looks strong in first click but weak in last click, it is probably driving top-of-funnel awareness. If it is strong in last click only, it is a conversion closer. Understanding these patterns helps you build a balanced media mix.
7. Setting Up Attribution in GA4 and Google Ads
GA4: Go to Admin > Attribution Settings. Choose your reporting attribution model (data-driven is the default since 2023). You can also set the lookback window for acquisition and all other events (7, 30, 60, or 90 days). For most ecommerce stores, 30 days for engaged events and 90 days for acquisition events is a reasonable starting point.
Google Ads: Go to Goals > Conversions > Settings > Attribution model. Each conversion action can have its own attribution model. For your primary purchase conversion, set it to data-driven (if you have enough data) or last click. Note that changing the attribution model retroactively changes historical data in some reports, so do not switch models back and forth frequently.
One important detail: GA4 and Google Ads can use different attribution models. Your GA4 reports might show one set of numbers and Google Ads might show another for the same campaigns. This is normal and expected. GA4 sees cross-platform journeys. Google Ads only sees interactions within Google. Compare, but do not expect them to match exactly. The discrepancies are actually informative, showing you how cross-platform journeys differ from within-platform ones.
For a deeper look at how attribution affects your specific ad accounts, our analytics and tracking services include a full attribution audit.
Frequently Asked Questions
Google Ads defaults to data-driven attribution for conversion actions that have enough data (300+ conversions in 30 days). For conversion actions with less data, it falls back to a model similar to last click. You can manually set the model for each conversion action.
GA4 sees the full cross-platform journey (Google, Meta, email, organic). Google Ads only sees interactions within Google. Also, they may use different attribution models and different lookback windows. A 10-20% difference is normal.
You can, but avoid doing it frequently. Each model change affects how historical data is displayed and how Smart Bidding optimizes. If you need to switch, do it between campaign cycles and give the algorithm 2-4 weeks to recalibrate.
See How Your Campaigns Really Perform
COREPPC's free audit shows your performance across attribution models so you can make better budget decisions.
Start Free Audit
