Beyond the last click: choosing an attribution model

Attribution models that fit your business

The path to a purchase keeps getting longer and more winding. Yet many advertisers still judge a campaign by the last thing the buyer saw “in the shop window.” In other words, the last ad click right before the purchase gets all the credit. In 2018, when we first wrote this article, it was the most widely used attribution model.

The problem is that the attribution model changes how you judge every campaign.

The path to a purchase can run through:

• several advertising channels (Search, Display, Facebook Ads campaigns, and so on);
• several devices (phone, tablet, laptop, and back again).

A carefully chosen attribution model gives each step the credit it deserves and shows you how much each part of your campaigns contributes.

What has changed since then: in 2023, Google Ads and Google Analytics 4 retired the first click, linear, time decay, and position-based models. What remains is data-driven attribution, now the default, and last click. Google explains the current options in its help article on attribution models. The retired models are described below because they’re still useful for understanding how attribution works, and you’ll still find them in other analytics tools.

Data-driven attribution (DDA)

The most accurate way to measure what each click contributes is data-driven attribution (DDA).

The model analyzes the paths to purchase in your account and splits the credit for each order among the clicks that contributed to it. In 2018, it needed a rich history (at least 400 conversions of the same type in the previous 28 days) and Analytics 360. Today, Google has made it available to almost all accounts, without those thresholds.

Its advantages:

• it’s simple to use: you don’t set the weights, the model calculates them;
• it assigns a value to every step toward the purchase;
• it works both for long, complex buying journeys and for very short ones with few steps.

The other models

The most restrictive models are first click and last click. They show you very little of the valuable interactions that happen after the first click or before the last one.

First click: the first ad click gets all the credit for the order

First clicks usually come from broad keywords, when someone is searching for a product or service without a preferred brand. This model suited advertisers running awareness campaigns for a new brand, for whom conversions were secondary for the time being.

Last click: the last ad click gets all the credit for the order

If the clicks before a purchase have no value for you, and your campaigns are strictly about sales volume with a short buying process, you can use this model.

Linear: every interaction gets equal credit

It lets you optimize campaigns taking into account all the clicks that led to the result.

Time decay: clicks closest to the purchase get the most credit

It gives value to every click along the way but emphasizes the ones that triggered the conversion.

Position-based: the first and last interactions get 40% of the credit each

The interactions in between share the remaining 20%. It suits advertisers who mainly want to track the clicks through which buyers discovered the brand and the clicks that led directly to the purchase.