Attribution models for an e-commerce

Do you really know what your marketing channels contribute?

When you manage an e-commerce you have multiple marketing channels in which the user has been passing before making a conversion to the website: Paid Search, Paid Social, Organic searches (SEO), organic Social, affiliation, programmatic, traffic Direct on the web, among others. So you need the Attribution Models To take control.

Users interact with the brand or business more than once until deciding to carry out the action that is expected of them. For this reason, with the attribution models they will help you to understand the way in which the marketing channels interact. It would not be fair that the last channel by which the user enters to perform the conversion takes the value completely (how is the last-interaction model), as they may have influenced other channels to achieve the goal. This is also another reason why to implement several channels and manage in unison, as they are SEO and SEM. Virtually all companies use more than one marketing channel.

So, if a user visits a website from different channels to meet the desired goal, it is important to understand how each channel participates in achieving that goal either a purchase, download or form and have a full view of what is happening in our marketing ecosystem.

Main types of attribution models

The following attribution models can be found in the Web analytics tools:

Linear Attribution Model

The tool establishes the credit or value equitably to each channel with which the client has interacted with the business. That is, if the client has seen my ad in Paid Search and later days buys me by entering directly the domain of the Web, assign the conversion in a 50% to each channel.

• Advantage: It is useful when in the purchase cycle or Customer Journey Of a user has been designed to maintain their contact and notoriety.
• Contra: does not show which channel has higher incomes or greater potential.

First Click attribution model

This model gives the full value of the conversion to the first channel that the user used, regardless of the scope of the subsequent channels.

• Advantage: It is very good to know the channels that initiate the interaction of the user with the brand.
• Cons: All customer Journey Contact points are not taken into account and therefore complicates optimization.

Last- Click Attribution model

It is the default model in many Web analytics tools. It grants all the credit of the conversion to the last channel with which the user interacted, regardless of the scope of the previous channels.

• Advantage: If all your efforts are to attract traffic to your website or sales, this model may be right.
• Cons: The last click may be one of the most important, but as in the previous model, it does not account for all channels prior to the conversion channel.

Time Impairment Attribution Model

It is an exponential model, that is to say, the channels closest to the conversion take more value and then the rest of the channels receive less value according to the time that passed since the last interaction of the user with that channel. And how is each value set? Well, seven days before the interaction that made the conversion, ie, the penultimate interaction, will receive half the value of the interaction that made the conversion. Then, fourteen days to the last interaction that made the conversion, you will receive a quarter of the conversion value. This model has a 30-day window.

• Advantage: It is ideal when you want to launch a promotion for a limited time, thus giving more value to the interactions close to the conversion in promotional period.
• Contra: Never going to give the right value to the marketing channels in the top of the funnel.

Attribution model according to position or U shaped attribution

It is a mixture of the models First Click And Last Click. It grants 40% to the first and last channel; The remaining 20% is distributed between the intermediate channels.

• Advantage: Usually the first and last points of contact are the most important, since the first ones are the first contact of the user with your brand while the last is the one that will take into account all the sales. Then, if these two features are important to your business, this can be your attribution model.
• Cons: It does not account for marketing efforts beyond achieving potential customer conversions.

Other types of attribution models

Indirect last Click Attribution model

Is the default model of Google analytics and, unlike the model Last Click, this will assign the entire value of the conversion to the last channel of interaction without taking into account the direct channel.

• Advantage: It can be useful when you think that direct traffic is from customers that you have already achieved by other channels.
• Contra: Your brand campaigns can be really affected.

Google Ads Last click Attribution model

This model can only be found in the Google Ads tool and is to grant the total conversion to the last ad in which the user clicked.

• Advantage: It can be useful to know which ads are the most relevant in terms of conversions.
• Cons: It does not account for the rest of the channels.

Logarithmic attribution model (Data driven)

Possibly the best model of attribution existing because it is the most accurate and adjusted to each business. The value assigned to each point of contact is attributed logarithmically through a matching learning of tool data. That’s why you need a lot of accumulated data. This model is available in Google Analytics 360 or SAS Customer Intelligence 360, among others.

• Advantage: It adapts to your business in a very efficient way to obtain a real value of each channel.
• Cons: The tools that have this model are high in price, plus it is a slow implementation process.

An example to try to take the theory to the practice of the main models of attribution is as follows:

“A user is hit with a Display ad when he read a blog about home decor, as he just moved to his new home and is looking for ideas to decorate. Click on the ad to get to know the brand and visit a couple of pages and get out. A week later he performs a Google search of “kitchen furniture”, as he intends to buy a board game and chairs, and click on the advertisement of our brand. See the products and get out. Although he has seen a model that he liked, he prefers to look at other competitors. Finally, days later he is impacted in his mail with a remarketing campaign with the product he was seeing of our brand and that he liked. I decide to make the purchase, enter through our domain and buy.»

Then the example customer journey scheme would be as follows:

Display > Paid Search > remarketing Direct

• Linear: The credit of the conversion between the four channels would be shared, so it would be 25% for each one.
• First Click: 100% of the credit of the conversion to the Display channel would be granted.
• Last Click: 100% of the credit of the conversion to the direct channel would be granted.
• Deterioration in time: In this case, the direct channel would receive the maximum credit of the conversion, Remarketing would receive half of the credit and so on as it moves away from the last channel of conversion.
• U shaped Attribution: a the Display and direct channels would be allocated 40% to each, as they are the channels that have opened and closed the sale, while Paid Search channels and remarketing would be granted 10% to each one.

Therefore, there is no better or worse attribution model, but if it is suitable for your business objectives. The first thing you need to know is how the user journey is, what contact points you want to measure, because the more you have, the more complex your optimization will be. You always want to be the most accurate and get the most real data, but you should also be aware of the knowledge and resources that you have.

It is very important that you know how each of them works to find out which one is ideal. If after reading this post you still do not have very clear what model of attribution is suitable for you, do not hesitate more… consult now!