Attribution Models in Google Analytics & Google Ads

An attribution model is a set of rules for sharing conversions between different traffic channels.

A conversion is any kind of beneficial action for your business taken by a customer, such as an order, a call, a sale, and so on.

A traffic channel is a source that leads users to your landing page. Usually, that means your business’s website, but apps or even quizzes could also act as landing pages.

Attribution models are crucial for analytics frameworks employed in tagging your landing pages, such as Google Analytics, Mixpanel.com, and others. They also see use on ad platforms integrated with social media, such as Facebook, Instagram, TikTok, and more.

The old “Rule of 7” states that a customer must encounter a service 5 to 7 times before they accept your offer. Regardless of how true that theory might actually be in cold, hard numbers, it’s still a nice illustration for why you need attribution models.

Here are some interactions a user might carry out before buying something in your online store:

1. They find the item via Google Shopping and visit your website to learn more.
2. They see a promoted post on social media, which prompts them to visit the website again.
3. They access the website via a newsletter.
4. They’ve been Googling relevant information and access the website via a search engine ad.
5. They Google the website directly, click the link, and add it to Favorites or bookmark it.
6. They access the website from their Favorites list or bookmarks and buy the item.

This adds up to six different traffic channels—Google Shopping, social media, newsletters, search engine ad, SEO, a direct click—that resulted in a single conversion (sale).

An analytics framework uses attribution models to assign conversion value to a specific channel, which allows a SEO consultant (business owner, marketing specialist, etc.) to select the most effective means of conversion.

Analytics frameworks come with their own settings and quirks, and they use different attribution models.

Here are the most common attribution models, used pretty much everywhere:

1. First interaction. The value is assigned to the traffic channel that initiated the interaction funnel leading to the eventual conversion.
2. Last interaction. This is the same as the above but for the last click in the funnel.
3. Last indirect interaction. Often set as the default model. Conversion is allocated to the last traffic channel in the funnel. If the last channel ends up being a direct click, then the second-to-last channel is selected instead.
4. Linear. The traffic channel value is spread evenly between all of the steps. For instance, if the user performed 5 actions before a conversion, then each channel is assigned a value of 0.2 conversions.
5. Time decay. The more recent actions are assigned more value, and vice versa.
6. Data-driven. Machine learning algorithms automatically assign conversion value to channels based on previous conversion and user action data.

Google Analytics uses the “last indirect interaction” model by default. The Multi-Channel Funnel Settings menu allows you to create any model based on an existing template.

You can use the Model Comparison Tool in the Conversions menu or the Attribution project to compare the performance of different models.

Conversion value is assigned to the last interaction channel.

Ignoring all the direct clicks, the conversion value is assigned to the last channel that led to the direct conversion instead.

The conversion value is assigned to the last click on a Google Ads ad before the conversion.

The value is assigned to the first interaction, regardless of its nature.

The conversion value is evenly spread among all the steps of the conversion funnel.

The more recent channels get assigned greater conversion value, and vice versa.

40% of the conversion value is assigned to the first and last interaction. The remaining 20% is spread evenly between all the remaining steps.

Learn more with this Google Help doc:

Attribution models for Multi-Channel Funnels

Multi-Channel Funnels

The “Attribution” project

The Google Ads service can be used as a standalone tool with no Google Analytics involved, which is why it comes with its own attribution models and settings to keep in mind.

Whereas Google Analytics spreads the conversion value between all of the traffic sources, Google Ads also accounts for all the ad campaigns, ads, and keywords used.

Thus, this is a model for users who have only interacted with Google Ads before a conversion occurred. The conversion value can be assigned to one of the Google Ads they clicked along the way, or it can be spread among all of the Google Ads within the funnel.

You can go to Settings > Attribution on the Google Ads dashboard to compare different models.

The attribution model is set up manually for each single conversion via the Conversion menu. To change it, click the desired conversion and then select “Change settings”.

The conversion value is assigned to the last clicked ad and keyword.

The conversion value is assigned to the first clicked ad and keyword.

The conversion value is spread evenly among all the ad interactions.

The more recent conversion steps receive greater conversion value, and vice versa.

The first and last interaction each receives 40% of the conversion value. The remaining 20% is spread evenly among all the other ads.

The Google algorithm assigns conversion value automatically using accumulated data for the selected conversion. However, this requires a certain data pool before it’s really useful. We recommend this model if you have enough data.Learn more with the Google Ads Model Attribution Help document.

Selecting the model that’s right for you is a crucial step,  with implications for your traffic analytics and many key business decisions. Choosing the wrong conversion model and disabling a valuable traffic source may break your conversion funnel entirely.

Before choosing your attribution model, think about the following questions:

1. Which conversion methods are the most crucial for your business?
2. How many steps does a user typically take before acting on a conversion?
3. Do you sell regular or unique/one-time products and services?
4. What is your end goal? Do you want to expand your audience, or are you focused on keeping your core clients and maintaining regular demand?
5. How much time does a user typically take before a conversion occurs?
6. Are you an established brand, or are you just breaking into the market?
7. Do you have enough accumulated conversion data, or are you just setting up your analytics pipeline?
8. How intense is the competition in your market?
9. Is your business affected by a certain time of the year or other irregular factors?

To come to the right decision, look through your Google Analytics reports, compare how different models perform in real time, and examine multi-channel funnels, associated conversions, as well as your own know-how and experience with your target audience.

Afterwards, try applying different attribution models to your pipeline and compare the resulting reports. Keep an eye out for cases that lead to a noticeable disparity in results.

The examples below will help you arrive at the right conclusions based on your existing data.

Exhibit A: A small barbershop with top-notch staff. You provide a regular service, and your clients stick with your business after visiting once. Your goal is expanding your clientele. The “first interaction” model works best here: without it, nobody would even know you exist. This first click is what seals the deal.

Exhibit B: An online store selling car tires. You’re in a competitive market selling a regular product that offers nothing particularly unique but is needed from time to time. The main value here would lie in the channel that convinced the user to buy their brand of car tire in your particular store, rather than anywhere else. Last indirect interaction would work best here.Exhibit C: Buying an apartment from a real estate developer. The funnel here involves much deliberation on the part of the buyer before they make their final choice, which is a large, one-time purchase. In this case, every traffic channel is very valuable, as each interaction exposes the user to your offer and gradually nudges them towards an important decision. The linear model would work best here.

Of course, these cases are merely examples of quick interaction funnel analysis, and they don’t make for a foolproof algorithm. Make sure to examine your business from all angles before changing your attribution model.

Attribution is a crucial parameter that defines the success of your ad campaign, as well as other key strategic decisions. Choosing the right model will greatly boost your traffic and make your ad campaigns that much more effective at reaching their target audience.