What is an attribution model?
An attribution model can be defined as a framework for measuring the contribution of each touchpoint in the conversion process. The idea is to be able to measure the ROI of each channel and adjust your budget accordingly.
To make the concept more tangible, you can compare an attribution model to a collective action in a soccer game, for example. When a goal is scored, we tend to only remember the player who made the decisive pass and the one who scored, but many players contributed earlier in the sequence. You could imagine that if the goalkeeper hadn't made a save and launched a good counterattack, the goal might never have happened. It therefore makes sense to credit part of the success to the goalkeeper.
Attribution models differ in how much credit they assign to each player who was involved in the sequence.
Why do we need these models?
The customer journey has grown significantly more complex in recent years, and exposure to ads and marketing channels has multiplied. It has become harder to determine how much each factor contributed to a conversion. Here are a few figures to illustrate this:
- We are exposed to 350 ads and more than 5,000 brands per day (source: Joanna Yakin, Radio France).
- The average internet user clicks on 4 ads on average before making a purchase (source: Just Search).
- 90% of consumers use multiple screens for their online activities (source: Just Search)
The period between the first touchpoint and the purchase is referred to as the customer journey:
Attribution models have therefore been developed and continue to evolve to establish precise rules for each touchpoint's contribution, align more closely with the customer journey, and better understand the factors that trigger conversions.
Here are the most common attribution models:
- First click
- Last click
- Position-based
- Linear
- Time decay
- Data-driven
- As mentioned earlier, the customer journey is growing increasingly complex. This is why first-click and last-click models are largely outdated, as they assign 100% of the conversion credit to a single touchpoint. Note that, by default, tracking tools such as Google Analytics are based on the last-click model.
- The linear model assigns equal credit to all channels without distinction.
- The position-based model considers that the first and last touchpoints played a greater role than the others in the conversion journey, and together they share 80% of the credit equally. The remaining 20% is distributed across the other channels.
- There is also the time decay model (ascending or descending). In one version, the credit assigned to channels increases with each touchpoint. In the other, it decreases with each new channel encountered.
- Finally, the most reliable yet most opaque model is the data-driven one. Here, the algorithm determines which channels have the greatest impact by taking into account a large number of factors such as user engagement with your ads, your account data, your keywords, and your business objectives.
This model therefore requires a certain volume of data to function properly. To be precise, you become eligible for this model when you record at least 3,000 clicks on Google.fr search pages in compatible networks, and a conversion action must have recorded at least 300 conversions over 30 days. The model may therefore be available for some conversions and not others. Once eligible, you must also maintain a data flow to keep benefiting from the model: at least 2,000 interactions with your ads in compatible networks or at least 200 conversions for the conversion action over a 30-day period.
It is also worth noting that this is the only adaptive model, meaning it is based on machine learning.
What impact does this have on Google Ads?
The attribution model only applies to clicks from the Google Search Network and the Shopping Network. No other channels are taken into account.
Within the platform itself, Google Ads applies the chosen attribution model. At the campaign, ad group, or keyword level, it will, for example, assign 1.6 conversions to one ad group and 0.4 to another.
The model you choose determines how values in the Conversions and All Conversions columns are calculated.
Since attribution models are based on conversions, your automated bidding strategies that rely on conversions will also be affected. The algorithm will optimize your bids accordingly.
To get a clearer sense of how your choice of model will impact results, you can compare models directly within the tool. Here is how: Tools > Insights > Attribution.
Once the Attribution tab is open, select "Model comparison" from the left-hand menu.
As you can see, you can choose the dimension you want to analyze (Campaign, Ad Group, or Keywords), the conversion action, the analysis period, and finally the two models to compare.
This model comparison feature is also available in Google Analytics.
In conclusion, attribution is essential for advertisers and agencies to:
- Better understand the different stages of the customer journey and reach customers earlier in the funnel.
- Optimize that journey from the first to the last step.
- Allocate budget to the most profitable channels for your industry.
All models have their merits, but the ideal is to work toward the data-driven model. It is the most comprehensive since it factors in a large number of variables and highlights what works best for your specific situation.
Note that you should not confuse an attribution model with an attribution window. The attribution window is the time period during which all or part of a conversion can be attributed to a touchpoint. The relevant duration varies by industry, ranging from a few days for everyday purchases to several weeks for B2B.
While Facebook recently reduced its attribution window from 28 days to 7 days, Google still allows you to define a longer period. You can configure this under Tools > Insights > Conversions.
Once the conversion is selected, simply update the settings and the conversion tracking period.
Finally, we want to let you know that Google Analytics has developed an Attribution feature that is still in beta. Its goal is to create attribution projects that provide the following insights:
- Accurate conversion tracking across all digital channels, with deduplication
- A consolidated and consistent view of all your digital performance
- A better understanding of the customer journey associated with your brand
Feel free to reach out if you would like to explore this topic further!
By Ronan Lecointre
.jpg)
.jpg)