Attribution Modeling for E-Commerce: Give Credit Where It's Due

The Attribution Problem

A customer sees your Facebook ad on Monday. They do not click. On Wednesday, they see your TikTok ad and click but do not buy. On Friday, they Google your store name and make a purchase.

Which channel gets credit for the sale? Facebook (first touch)? TikTok (assisted)? Google (last touch)? All three? This is the attribution problem, and how you answer it determines where you spend your marketing budget.

Why Attribution Matters

If you use last-click attribution (the default in most analytics tools), Google gets 100% credit for that sale. Your data shows Google as a profitable channel and Facebook and TikTok as money pits generating clicks but no sales.

Based on that data, you might cut Facebook and TikTok budgets. But those channels were driving awareness that led to the Google search. Without them, the sale never happens. Wrong attribution leads to wrong decisions.

Common Attribution Models

Last Click

How it works: 100% credit goes to the last channel the customer interacted with before purchasing.

Pros: Simple, easy to understand, default in most tools.

Cons: Ignores all channels that contributed to awareness and consideration. Overvalues brand search and direct traffic. Undervalues awareness channels like social ads.

Best for: Stores with simple, single-touchpoint customer journeys.

First Click

How it works: 100% credit goes to the first channel the customer interacted with.

Pros: Highlights which channels are best at introducing new customers to your brand.

Cons: Ignores everything that happened between first touch and purchase. Overvalues awareness channels and undervalues conversion channels.

Best for: Understanding which channels drive new customer acquisition.

Linear

How it works: Credit is split equally across all touchpoints in the customer journey.

Pros: Acknowledges every channel's contribution. No channel is ignored.

Cons: Treats all touchpoints as equally important, which is rarely true. A casual ad impression is not as valuable as the final purchase click.

Best for: Stores wanting a balanced view of all channels.

Time Decay

How it works: More credit goes to touchpoints closer to the purchase. Earlier touchpoints get less credit.

Pros: Acknowledges the full journey while giving more weight to the channels that sealed the deal.

Cons: Can still undervalue awareness channels that planted the seed days or weeks earlier.

Best for: Stores with longer purchase decision cycles.

Position-Based (U-Shaped)

How it works: 40% credit to the first touchpoint, 40% to the last touchpoint, and 20% split among middle touchpoints.

Pros: Highlights both the channel that created awareness and the channel that closed the sale. Acknowledges assist channels.

Cons: The 40/40/20 split is arbitrary and may not reflect reality.

Best for: Stores wanting to value both acquisition and conversion equally.

Data-Driven

How it works: Uses machine learning to analyze all conversion paths and assign credit based on actual contribution patterns.

Pros: Most accurate model. Based on your actual data rather than arbitrary rules.

Cons: Requires significant data volume (typically 600+ conversions per month). Available in GA4 and some advanced platforms.

Best for: Stores with enough data volume to support machine learning models.

GA4 Attribution Settings

GA4 uses data-driven attribution by default if you have enough data, and falls back to last-click if you do not.

To check or change your attribution model:

Go to Admin > Attribution Settings
Select your preferred reporting attribution model
Choose your conversion lookback window (7, 30, or 90 days)

Recommendation: Use data-driven if available. If not, use position-based (U-shaped) for a more balanced view than last-click.

Platform-Reported vs. Actual Conversions

Every ad platform reports its own conversions, and they all take as much credit as possible:

Facebook reports: 50 purchases attributed to Facebook ads
TikTok reports: 30 purchases attributed to TikTok ads
Google reports: 25 purchases attributed to Google ads
Your Stripe dashboard shows: 60 total purchases

The platforms collectively claim 105 purchases, but only 60 actually happened. This is because multiple platforms claim credit for the same purchase.

How to handle this:

Use your payment processor (Stripe) as the source of truth for total revenue
Use platform reporting for relative performance within each platform
Use GA4 for cross-channel attribution
Accept that attribution will never be 100% accurate

Practical Attribution for Small Stores

If you are spending less than $5,000 per month on ads, complex attribution modeling is overkill. Here is a practical approach:

1. Track Total ROAS

Divide total Stripe revenue by total ad spend across all platforms. This is your blended ROAS. If it is profitable, your overall marketing mix is working.

2. Use Platform Metrics for Within-Platform Decisions

Use Facebook's reported data to decide which Facebook campaigns to scale or kill. Use TikTok's data for TikTok decisions. Do not compare platform-reported metrics across platforms.

3. Run Channel Isolation Tests

The most reliable way to measure a channel's true impact: turn it off for a week and see what happens to total revenue.

Pause Facebook ads for one week. Did Stripe revenue drop? By how much?
Resume Facebook, pause TikTok. Compare.

This is crude but effective. If pausing Facebook drops total revenue by 40%, Facebook is responsible for roughly 40% of your revenue regardless of what attribution models say.

4. Post-Purchase Surveys

Add a "How did you hear about us?" question after purchase. This self-reported attribution has biases (people forget or misattribute) but provides a useful signal.

The Attribution Mindset

Perfect attribution is impossible. No model captures every interaction across every device over every time period. The goal is not perfection but directional accuracy.

Ask yourself: "Does this attribution model help me make better budget decisions than guessing?" If yes, it is useful. Do not chase perfection at the expense of action.

Key Takeaways

Last-click attribution undervalues awareness channels and leads to under-investing in top-of-funnel marketing
Every ad platform over-reports its conversions so use Stripe as your source of truth for total revenue
Position-based attribution gives a more balanced view for most stores
Data-driven attribution in GA4 is the most accurate but requires 600+ monthly conversions
Blended ROAS (total revenue / total ad spend) is the simplest and often most useful metric
Channel isolation tests provide the most reliable real-world attribution data
Perfect attribution is impossible so focus on directionally correct decisions rather than precise numbers