Attribution Modeling in Campaign Manager 360: From Beginner to Pro

It’s important to take a holistic approach to marketing and utilize several channels to attract new traffic and increase conversion rates. However, some issues still arise:

How to conduct an assessment of your marketing campaigns' effectiveness?
How to ensure data integrity?
How to process large data volumes?

The answer is obvious: you need to analyze attributed data correctly.

Anton Logonovich, Google Marketing Platform products coordinator at Admixer Advertising, will help you figure out what attribution is, what types of attribution models exist, what kind of data can be extracted, and the future of attribution after the cancelation of third-party cookies.

What is Attribution?

Attribution is the process of distributing the conversion value between all user touchpoints with your ads. With its help, you can identify which campaigns and creatives on different platforms made the user perform the targeted action. Attribution helps you optimize your strategies and make data-driven decisions, improving advertising effectiveness and ROAS metrics.

Every marketer and advertiser wants their prospects to convert immediately after they see an ad. Unfortunately, in reality, this process is far more complicated. Users may need to interact with an advertisement several times before making a purchase decision. For example, a prospect might search for a product in Google, open your website, then leave without purchasing. And after some time, they might begin to notice paid ads from your e-commerce platform, which prompts them to click through and purchase your product. By this point, how will you know which of the channels truly led to the conversion?

This is where attribution comes to the rescue. By selecting the right model, you’ll understand how to allocate your marketing budget to achieve better results.

Attribution Models

An attribution model focuses on the distribution of lead value between interaction channels. It allows us to identify the role and importance of each channel in the consumer’s decision-making process.

Let’s consider the following example. Without PPC advertising, users probably wouldn’t enter the company’s website. However, without remarketing, the purchase would not take place. It appears that each of these interactions has its own value, and a thorough analysis of them will help you identify the effectiveness of your advertising campaign and the ROMI of each channel for each campaign.

Google Campaign Manager 360 offers the following attribution models.

Floodlight

Here, the conversion value is counted to the user’s last touchpoint with the advertisement before completing the targeted action. What’s more, the click takes precedence in this case. For instance, on Monday, the user clicked on the paid search ad, and on Tuesday, they saw a pre-roll on YouTube. This attribution model counts the conversion on the paid search ad click.

Last Interaction

This model assigns the conversion value to the last channel in the interaction chain. In the previous example, that would be the YouTube pre-roll on Tuesday.

First Interaction

This model assigns the conversion to the first contact with the ad in the chain. In our example, that’s the click on the paid search ad.

Linear

In this case, all touchpoints in our chain have a similar value. For instance, if the user clicks the ad on Monday and sees the pre-roll on Tuesday, the system will assign a 0.5 conversion value to the click and 0.5 to the YouTube pre-roll.

This model can be used in cases where you need continuous advertising contact with the user and every moment of the interaction is important for the decision-making process.

Time Decay

The main principle of the Time Decay model is: the closer the user interaction moments are to the conversion, the more valuable they are. The downturn period stands at seven days. For instance, if a user clicks on the ad on Monday and makes a purchase on Sunday, their click will be half as valuable as any interaction on the day of the purchase. What’s more, if it has been two weeks since the first click, the conversion value will be four times lower.

This attribution model is the most efficient for short-term marketing campaigns or cases where it’s important to count the last interaction while also keeping others in mind.

Position-Based

This attribution model combines the First Interaction and Last Interaction models, but instead of assigning a conversion value to one of the options, the system shares it between them. In most cases, the first and last interactions get 40% each, and everything between them gets the remaining 20%.

This model is perfect for those who treat interactions that bring users to the brand and channels that result in conversions equally.

Data-Driven Attribution (DDA)

The final model doesn’t take into account the marketing channel’s place in the interaction chain, but it analyzes and assesses the specific contribution of each contact to the conversion.

All of the models described above assign a key value to a certain source position, but it isn’t necessarily the one that brings the most conversions. The value is assigned based on the position of a channel in the interaction chain, but it doesn’t guarantee that the interaction actually influenced the conversion. It simply worked in a certain moment on a path to conversion.

On the other hand, DDA takes into account the diversity of paths to the conversion and assigns value to each user interaction. The degree of value depends on how a certain interaction influenced the final result rather than the place of the interaction in the chain. Furthermore, the model constantly trains using data refreshing.

It's best to opt for this model when you need to identify with utmost accuracy which channels and ads have the highest efficiency so you can reasonably distribute your budget. Also, it may come in handy if you’re working with large data sets, where it’s difficult to determine the actual impact of each channel.

Furthermore, you can create a custom attribution model using one of the standard models as a basis and supplementing it with the required parameters that are vital to your specific business.

Attribution Data in Campaign Manager 360

You can gather marketing data from YouTube, Search Ads 360, Google Ads, and Display & Video 360, as well as track various programmatic systems. You can also track clicks on Facebook. However, you can’t get access to metrics like reach, post-view conversions, and geolocation statistics with Campaign Manager 360.

Why Use Campaign Manager 360?

Let’s take a look at another example. Assume we have two search keywords: “Buy iPhone 12” and “Buy iPhone 11”. The first user clicked on the ad with the first keyword and made a purchase. The second user clicked on the ad with the keyword “Buy iPhone 11”, proceeded to the website, didn’t make a conversion, and, a few days later, bought the product via organic search.

If you use Google Ads, there are two conversion sources: Google Analytics goals and conversion tags in Google Ads. Google Analytics will count the conversion on the keyword “Buy iPhone 12”. However, it won’t count the keyword “Buy iPhone 11” because the system assigns this conversion to organic search. Google Ads algorithms will notice that “Buy iPhone 11” is performing poorly and the bids for this keyword should be lowered. However, that’s not true.

If you take a look at conversion tags in Google Ads, all keywords will have conversions assigned to them.

The real issues begin when media advertising and Facebook are added to the system. Most likely, you’ll lose control of the conversion attribution between these paid channels. Campaign Manager 360 helps to pull all of these pieces together, giving you an accurate view of what’s really happening. Media advertising and performance marketing data are transferred to Campaign Manager 360, where you can analyze all of the results in one place.

What Data can be Extracted From Campaign Manager 360?

Now, let’s take a look at the data that can be gathered from Campaign Manager 360.

Conversion Chains

Conversion chains data is information about the channels that users interacted with before making a conversion. The system allows users to generate reports that include user interactions (clicks or impressions) with Facebook, search, media ads, and the number of impressions. Later, these data can be streamlined to business intelligence tools for thorough analysis and building user behavior patterns.

The Number of Interactions Required for the Conversion

This data shows how many interactions users needed to make a conversion. This information includes impressions, clicks, or both. Knowing the reach, the number of impressions, and the advertising price, you can calculate the CPA for each channel and understand which channel is the most relevant.

The Number of Associated Conversions

This is the number of interactions with traffic sources that weren’t the last in the conversion chain but helped the user make the conversion. For instance, a person saw a pre-roll and an ad banner on the website but purchased via the Facebook ad. The last channel gets the conversion, while the remaining two will be classified as an associated conversion.

The Interval Between the First Contact and a Conversion

The interval between the first user interaction and the moment of the conversion allows you to identify the optimal time for the user to make a purchase.

The Future of Attribution After the End of Third-Party Cookies

Today, attribution accuracy is approximately 70%. There’s a theory that after the end of third-party cookies, this rate may decrease. However, Google has developed a new solution and is implementing conversion modeling.

'I don’t have access to Google’s documentation. That’s why it’s impossible to say exactly how this mechanism will work. But I think algorithms will assess the conversion time, draw parallels between displaying ads in different channels, then compare devices and other parameters. Based on the gathered data, the system will be able to model conversions. Due to this pattern, the attribution accuracy may stay on the same level or even rise. First of all, Google products have enough data without third-party cookies. Secondly, we don’t have enough information about Firefox and Safari users. Conversion modeling may fill in this gap.' — Anton Logonovich, GMP products coordinator at Admixer Advertising.