Attribution Models for Ecommerce

My company helps merchants analyze and optimize marketing data. Clients’ most frequent questions involve attribution. What’s the source of truth? What drove the purchase? What prompted the visit to my site?

Let’s start with attribution tracking in Google Analytics.

Google Analytics

Google Analytics 4 now offers just two methods for attributing conversions:

• “Data-driven” uses machine learning to distribute attribution across multiple sources based on users’ previous behavior, excluding direct traffic, although it appears to skew toward Google-owned channels.

GA4 offers multiple attribution windows, depending on a business’s sales cycle. Some products require no research and are typically purchased in minutes. Others are complex and need much consideration. I typically set the window at 30, 60, or 90 days.

Rarely does an ecommerce platform’s conversion attribution reports match Google Analytics. Here’s why.

• Technical errors, such as incorrect installation of pixels on Google or Meta ads, and mistakes with UTM parameters.

• Privacy rules and regulations complicate tracking. Examples include the E.U.’s GDPR and cookie restrictions.

• Non-digital promotions, such as ads on TV, print, radio, and billboards, do not appear in GA4.

• Multiple touches. A consumer may see a product or brand offline, search for it on Google, click on a paid listing, and then abandon the journey. Later, the product may appear in the shopper’s Instagram feed, prompting the conversion. No attribution scenario can pinpoint the source(s), as it varies by shopper.

• Repeat purchases. Some returning customers go directly to a website, while others respond to ads.

Despite the differences, Google Analytics remains the most-used attribution tool. It’s free, with an ecosystem of users, consultants, and resources. It’s a good choice for advertisers on Google-owned platforms, although it also captures referrals from other sources.

Other Methods

Still, merchants have other attribution options.

Ecommerce platforms. Shopify, for example, offers multiple attribution models — last click, last non-direct click, and first click — and multiple windows. Most platforms, including Shopify, show just one source per sale. Merchants with few marketing channels and single touchpoints can usually rely on their platform’s reporting.

Third-party tools. Segment, Adobe Analytics, and others utilize regression models for multi-touch attribution, similar to GA4’s Data-driven method of assigning a value to each source by channel or campaign. Third-party tools do the math but cost money. They are not as accurate as one would hope, in my experience.

Marketing platforms. Most marketing channels offer built-in reporting for performance tracking on that platform. Advertisers can monitor, for example, the creative, body text, and audience targeting. But in-platform reports are not ideal when contrasting, say, Google versus Meta.

Simplified approach. An easy-to-implement method is to compare daily sales from your ecommerce platform with GA4’s Data-Driven conversion attribution reports. Then assess GA4’s values to establish the source of truth. Apply over- or under-reporting in GA4 as a percentage to arrive at a return on investment per channel. Perhaps a TV ad or a brand campaign generated a sales boost. Neither would appear in GA4. While not exact, this simplified approach can provide a more accurate reflection of a channel’s impact on revenue.

Here’s an example. My firm just analyzed sales attributions for an ecommerce health food client. We found (i) a strong sales correlation with both Google Ads and email marketing, (ii) a moderate correlation with Instagram ads, and (iii) a weak to non-existent correlation with sales and TikTok Ads. However, we did see success with retargeting ads on TikTok.

No Perfect Model

I know of no perfect conversion attribution platform or technique. The purchase journeys of modern shoppers are too complex and varied. But we can consistently gauge the impact of a channel or campaign by establishing the right process for a merchant’s products, marketing tactics, and tech setup.