Which attribution model to choose?

Recently, the user path on the Internet has changed a lot. Before we make a decision to make a purchase, sign up for a training course or buy a vacation, we browse competitors’ sites, read reviews, go to the site with the chosen product, many times from different sources. When we analyze the path of our customer, we can see that the so-called “points of contact” with our brand were many before the conversion occurred. On the other hand, the conversion of such a customer itself is attributed differently to different traffic sources. How to properly attribute conversion data and how to analyze it? You can find out in the following article.

WHAT IS ATTRIBUTION?

Attribution, which will be discussed today, is precisely the attribution of conversions to a specific traffic source. Attribution models, on the other hand, allow you to determine how much of a particular interaction with an ad will be considered a conversion source. This gives you the ability to get more valuable information about the effectiveness of your ads and optimize conversion paths.

Why is it worth using ATTRIBUTION MODELS?

We tend to look at our advertising efforts through the lens of the last click. So we attribute the entire share of conversions to the last click and its associated keyword. By doing so, we overlook many other interactions with the ad that may also have determined the final conversion. Using attribution models gives us more control over what share of the conversion will be attributed to the various interactions in our user’s path. With this knowledge, we can, among other things:

• reach our potential customers at the very beginning of their purchase path,
• select the right attribution model for our business so that it best reflects the conversions along our potential customer’s path
• with more accurate information about ad interactions, we can set rates more precisely and effectively

MODELS OF ATTRIBUTION IN GOOGLE ADS 

In the Google Ads panel, we can use various attribution models:

1. Last Click (Last Click) – the entire share of conversion is attributed to the last source of visit to our site. 
2. First Click (English: First Click) – the total share of conversion is attributed to the first source of interaction. 
3. Linear/linear (English Linear) – all interactions with an ad are assigned an equal share of the conversion.
4. Share decay over time (ang. Time decay) – in this case, the largest share of conversion is attributed to clicks that occurred in the shortest time before the conversion was realized. In this attribution model, we must remember that a period of 7 days is counted from the time of interaction with the ad. This means that if the interaction with the ad occurs on the 8th day, it receives half the share of the conversion than if it occurred one day before the conversion.
5. Position-based (Position-based) – in this attribution model, 40% of the conversion share is attributed to the first and last source of interaction with an ad and the keywords associated with that ad. The remaining 20% is distributed evenly among the remaining interactions along the path.
6. Data-driven (Data-driven) – this model analyzes and compares all interactions along the conversion path. As a result, those campaigns, ad groups and keywords that occur most frequently on a user’s conversion path receive the most value. An additional benefit that comes from this attribution model is that it analyzes not only the paths that led to success, but also those where users did not convert. By doing so, the system will learn equally valuable data. It will know which actions increase the likelihood of a user making a conversion. 

When using the data-driven attribution model, it is worth remembering that it is not immediately available, and especially not in a new Google Ads account. To use it, we need to collect a minimum of 3,000 clicks and 300 conversions in the last 30 days. 

ATTRIBUTION MODELS – HOW TO ANALYZE CONVERSIONS?

You already know that assigning a specific attribution model to a conversion allows you to analyze what share of the conversion a particular action had. Setting an attribution model also affects conversion counts in the Google Ads panel. We will see the differences in the “Conversions” and “All conversions” columns. How do they differ from each other? The “Conversions” column shows us all the actions that we set as “main” while the “All conversions” column will also show us the actions marked as “additional”. If you choose the “last click” attribution model, the data you will see in the conversions column will present those that were the user’s last click before the conversion was realized. 

You may also encounter conversion data presented in decimal numbers, but this is most common with item-based and data-driven attribution.

WHERE TO ANALYZE MODELS OF ATTRIBUTION? 

RAPORT: COMPARISON OF ATTRIBUTION MODELS

Now that we know how the different models work, and the conversions we set up collect data with attribution models assigned to them, we would like to be able to analyze and compare the collected information. Where can we find this?”

In the Google Ads dashboard after clicking on “Tools and Settings” → “Effectiveness Measurement” → “Attribution“. On the left side of the panel, you will find the option “Model Comparison“, and there you will find the attribution report and all the most important information for you. 

First, it’s a good idea to select the specific date range we’re interested in (we also have the option to compare selected periods to each other). Next, we can decide whether we want to compare campaigns alone, or expand our report to include ad groups, keywords or even the device from which conversions are recorded. If we have many conversions added to our account, and we want to focus on one selected one – just expand the field titled “Action causing conversion“. Next, we select from the list the validity period – 30, 60 or 90 days (that is, the length of the pre-conversion period from which ad displays are taken into account). 

In the last step we go to the heart of – we choose which two attribution models we want to compare with each other. Once this step is completed, the table will update with the criteria we selected. What conclusions can be drawn from such a report? Take a look at the example.

• If we compare the “First click” and “Last click” attribution model with each other, we can see which keywords users start the conversion path with. 
• By comparing the “Last Click” vs. the “Linear” attribution model, we will find the keywords that are used most frequently throughout the user path.
• On the other hand, when we compare “Last click” vs. the “Data-driven” model, we get information about the value of these keywords as determined by Google’s self-learning systems.
• Analyze CPA or ROAS by looking at different attribution models and using the columns “Conv. cost” and “Conv. value. / cost.” Such analysis will later help you adjust rates based on the actual value of the campaign and keywords.
• It is still worth comparing the “Last Click” model with the “Data-Based” model and look at it in the context of the campaign. If we see big differences between the models, it will give us valuable information that this particular campaign contributes not only to the user’s last step, but at different stages of the conversion path. 

RAPORT: SUPPORTED CONVERSIONS

What can we learn from this report? We will extract information on how many conversions our campaign, ad group, keyword, creative or device had a part in earlier than with only the last click. 

What else important can we see in this report? Which dimensions, at what stage of the user path yield the best results. Using the “supporting conversions/conversions coming from last click” column, we can see whether the selected campaign closes the user path sooner (score less than 1), opens it (score greater than 1) or participates both at the beginning of our user’s path and supports the other campaigns (score 1 or very close to 1).

RAPORT: PATH DATA

This is another report available under the “Attribution” tab. It shows how many days and interactions on average must pass for our potential customer to convert. Once we know how much average time users need to think, we can use this information when planning remarketing activities. 

Report: THE MOST IMPORTANT WAYS

In this report we will find information from which dimensions (campaigns, ad groups, keywords, creatives, devices) the interactions came from before the user converted. We will also see the length of such a path, as it was not always an interaction with one campaign. Sometimes a user had to go through three different campaigns and only after clicking on the last one – did he make a conversion. It is also worth tracking traffic from different devices at this point to see how and how many times during the path our potential customer moved between devices. 

WHAT MODEL OF ATTRIBUTION SHOULD YOU CHOOSE?

We will use here the answer least liked by everyone, but very true when it comes to the topic of attribution. Namely – it depends 🙂  

Depends on our business, our strategy and the activities we are currently doing. A different attribution model will work well for industry-specific queries, another if we want to analyze in detail how a user moves between steps in our online store.

As a summary of the topic, I’ll give you some examples in which it’s worth considering this rather than another attribution model:

• Last-click model: will work well for short, one-day promotions
• First-click model: in case you enter the market with a new brand or start, for example, cooperation with influencers 
• Linear – we can use this when we sell products that require a long lead time, as all interactions before purchase will be important to us
• Decrease in share over time – due to its specificity, we can use this attribution model for example for remarketing campaigns 
• Position consideration – it is worth using this attribution model when we have a whole strategy and sales funnel planned. What will be important for us will be the actions that opened the path and those that closed it. 
• Data-driven model – we will use in situations when, firstly, we have enough collected data in the Google Ads account, and secondly, we care about maximum effectiveness and optimization of our activities It is still worth mentioning here that the data-driven attribution model is the default model in the latest Google Analytics 4

Using the right attribution models in Google Ads will allow you to more accurately track the impact of different traffic sources on your conversions and make better advertising decisions!

Our knowledge and experience in the field will allow us to tailor attribution strategies to your business, giving you a better understanding of the effectiveness of your ad campaigns. Contact us to learn more and start effective Google Ads campaigns.

Let’s talk!

Now that you know what attribution models are and how to analyze them, it will be easier for you not only in your work with customers, but also in your work with your team in agency You’ll look more consciously at the data in the column table, making your conclusions for further action more relevant and effective. 

Good luck! 🙂