Google Analytics Attribution Model
Here’s the translated content with updated URLs:
How can you determine which marketing channels are more effective than others? Attribution calculates this — it helps to distribute value among all interactions with the consumer before a conversion: purchase, order, download, or something else. In this article, we will explain the differences between models and which ones consider each interaction individually.
Last Interaction Attribution
The most familiar model, which emerged among the first, and quickly became popular.
When to use: When there is established demand for a product, that is, in campaigns that attract customers at the point of purchase.
Example: Summer is always a low season for a gift store. You need to catch customers when they are ready to buy and actively searching for products. Therefore, focus on high-conversion queries like “buy a gift,” “gift shop,” “birthday gift ideas.”
Advantage: Optimization based on traffic that leads to conversion — targeting high-conversion and branded queries.
Disadvantages:
- High competition in the lower part of the funnel — competitors bid on the same queries.
- Formats aimed at a wide audience are not considered: display ads, video, and others.
- Overvaluation of direct traffic. Many users visit the site directly and buy products immediately. As a result, channels that initially played a role in attracting users suffer — before the direct visit.
First Click Attribution
100% of the conversion value is assigned to the first channel in the interaction chain.
This model, on the other hand, focuses on the first channel that introduced the customer to the product.
When to use: During the user’s initial steps toward purchase, when they are still undecided about whether they need the product, and if so, which one.
Example: Pre-holiday season — always a high season for a toy store. Everyone needs Christmas gifts, so attracting users at the initial stage of product search is essential. During this period, it may also be effective to purchase broad, but low-conversion general queries, such as “gift.” For a financial institution during a similar high-demand period, this might include terms like “loan,” “real estate,” or “car.”
Advantages:
- Optimization at the beginning of the user journey to purchase, when they still have not made a decision.
- Optimization for broad formats — display, video.
Disadvantages:
- Bias towards a single channel: all value is assigned to just one source.
- Subjectivity of the model.
Linear Attribution Model
All channels are equally valuable.
“No firsts or last” — this is the principle of the linear model, where weight is evenly distributed among all interactions. The principle is rarely used because it does not provide insights for optimizing the advertising budget. From this model, it is impossible to understand when the decisive interaction occurs.
When to use: If you need constant advertising contact with the user and value every touchpoint during the decision-making process.
Example: A year-round food delivery campaign. You need to maintain awareness among potential customers through broad formats while also capturing established demand with terms like “buy” and “order.”
Advantage: A more advanced model than the previous ones — considers all interactions.
Disadvantages:
- Presents an unrealistic picture, as all channels contribute differently to conversion.
- Roles of channels in user acquisition are also different — some recommend products, others convert clients. There are no equal roles.
Time Decay Attribution
The greatest value is assigned to interactions closest to the conversion.
The “main character” of this model is time. The further from the conversion, the less weight.
When to use: If it’s important to consider the last interaction but not lose other interactions. Relevant for short-term campaigns where interactions during peak days of the promotion are assigned the highest value, while interactions occurring 7 days before are less important.
Example: A peanut butter brand is running a contest “show unconventional use of the product on social media and get 3 free boxes.” The contest lasts through November, with the deadline for entries at the end of the month. It makes sense to assign the highest conversion weight to this period.
Advantage: Conversion weight does not entirely shift from broad formats to converting ones. Display, video, social media — all will receive a share of the conversion weight.
Disadvantage: Channels that first introduced the user to the brand are undervalued — they have unjustifiably low weight.
Position-Based Attribution
The greatest value is assigned to two interactions — the first (40%) and the last (40%). The remaining 20% is evenly distributed among the others.
When to use: If the interactions through which the consumer first learned about the brand are as important as the channels that led to conversion.
Example: A major consumer brand launches a new product. It is important to build product awareness from the start (through broad formats) and capture already established demand at the end (conversion queries).
Advantage: Highlights two most important points — entry into the funnel (broad) and the final converting interaction.
Disadvantage: Assisting interactions, i.e., those that helped the consumer reach the end, are undervalued — even if these points made a significant contribution to the conversion, they will have low value.
Last Indirect Click Attribution
The value is assigned to the last significant channel in the chain that occurred before all direct visits.
By default, this model is often used in many analytics tools. Some terminology:
- Indirect Click — a visit to the site from any channel (search, social media, video ads), including from a link on a third-party site.
- Direct Click — a visit to the site when the user types the address directly or navigates from a bookmark in the browser.
Thus, the highest value is assigned to the click that occurred before the direct one, when the person does everything themselves.
The chain may look like this:
Google CPC → Organic → Direct → Direct → Direct → Conversion
Another type of last indirect click attribution is “Last Click in Google Ads.” It is used when you need to assess the contribution of this specific channel. The chain of clicks might look like this (the main slice of the “conversion pie” will go to Google CPC):
Display → Google CPC → Organic → Referral → Social → Direct → Conversion
When to use: If you are buying traffic in Google Ads and need to identify the most converting formats and ads.
When direct visits bring back previously acquired clients — they no longer need to be considered.
Example: You are a local cheese brand already known in the market. You do not need to build brand awareness or find potential clients, but you need to capture all established demand and bring it to your site. Therefore, you need to optimize Google Ads and choose the most converting formats and ads.
Advantage: Determines Google Ads ads that brought the most conversions.
Disadvantage: All other sources in the customer interaction chain and their contribution to conversion are undervalued.
Data-Driven Attribution
Does not consider the position of the channel in the chain but evaluates the real contribution of each interaction to the conversion.
One of the main approaches was developed by Lloyd Shapley — a Nobel laureate who created a tool for distributing winnings among team members. In our case, the team members are interactions between the user and advertising channels, and the prize is the conversion. It is not about when a team member joined others but the specific contribution they made (influenced by many factors). Therefore, algorithms analyze interactions and determine the weight of each one.
When to use: When you need to accurately determine which channels and keywords are most effective and allocate funds as reasonably as possible.
Example: You are a car dealer, and your product is not bought immediately; people need time to choose a car and decide to buy. Therefore, you cannot rely on the last interaction — it will not happen without the others. You need to understand which channel truly impacted the conversion and which one can be removed without affecting the result. This is the “magic” of data-driven attribution.
Advantages:
- Allows for bid optimization.
- Not subjective and based on real data.
- Shows which channels and keywords are the most effective.
Disadvantage: Requires a lot of data for the algorithms to work correctly, so it is not available to all advertisers.
Read the article: “Data-Driven Attribution: What It Is and How It Works“
Custom Attribution Model
The model is built based on your own observations
The model is based on one of the standard models and supplemented with parameters important for your business.
When to use: If you have already tried standard attribution models, understood what they lack, and know which metrics are important for you to consider.
Example: You have been running a language course campaign for several years. You have conducted numerous tests, turning advertising sources on and off. You noticed that video ads had a significant impact on the success of the campaign. Therefore, you need to build an attribution model in which the highest value is assigned to the video format. To do this, you create a custom attribution model based on a basic model (e.g., position-based) and assign a value multiplier to the “video” channel.
Advantage: Customizable based on business needs.
Disadvantage: Requires a lot of preparatory work.
Attribution in Google: Tools and Data
Google Analytics
Attribution reports are available in the “Conversions” section. Advertisers have access to all described models, but the default reports use “last indirect interaction.”
Google Ads
Attribution reports are located in the “Tools” section — under “Search Attribution.” All types of attribution are available for modeling, and standard reports are based on “last click in Google Ads.”
Google Attribution (Beta)
The new system, which is currently in beta, will integrate attribution reports from Google Analytics. All attribution models will be available for comparative analysis.
Google Marketing Platform
Used to automate and optimize programmatic ad buying. If ads are bought on non-Google accounts, they can be tagged with pixels in the Google Marketing Platform.
