An attribution model is a set of rules that define which online channels get credit for a sale and/or conversion.

Indeed, unlike a few years ago when digital did not play a role in the purchase decision, today the customer journey is becoming more complex.  

A consumer now has many points of contact with your company before making a purchase. Moreover, as competition is tough and offers are plentiful, they need time and information before making a decision. They will search, compare, ask for advice and why not even wait for promotional offers. He can multiply interactions before going from prospect to customer.

And this is where multi-channel marketing was born.

A person may come across an advertisement on a social network, go to the company's website and sign up for the newsletter. A few days later, he will receive an email with offers, he clicks on it and goes to the site. He is retargeted in the following days and finally returns to the site to make his purchase.

To which channel do you owe the sale? To the advertisement that caught their attention and subsequently retargeted them? To the newsletters sent to inform the prospect of your offers and not to let him forget you? To the website on which they made their purchase?

You need to be able to define which channels have been useful to convert and to what extent, based on your objectives.

You will be able to analyze the effectiveness of your campaigns and choose to value one channel rather than another according to the habits and the Customer Journey of your audience.

The choice of the attribution model and its analysis will lead you to adjust your marketing according to the results obtained and your objectives.

I - SINGLE TOUCH MODELS

A - FIRST CLICK

The "First Click" or "First Interaction" attribution gives credit for the conversion entirely to the first channel used by the customer.

B2B example: If for example a prospect comes across one of your YouTube videos, goes to your site, you retarget him with a Google Ads ad and he returns to your site to subscribe to your offer, it is YouTube that will count the entire conversion.

‍

B2C example: If a consumer clicks on a Google ad and completes their purchase online, the sale will be attributed to the Google Ads publication.

It is the first channel with which the customer interacts that wins the day.

The advantage of this strategy is to be able to measure the performance of a campaign carried out to make a company known, or to know where the new leads come from.

The downside is that a single channel takes all the credit for a sale when multiple touch points were needed to convert the prospect into a customer.

This model is used to highlight that the activation of generic keywords is beneficial. The model is certainly not optimal but it allows you, especially in your Google Ads investments, to realize that these keywords are indispensable. Other attribution models would not allow you to realize their importance, and you could decide to stop using them when they are a key in your conversion cycle.

B - LAST CLICK

Unlike the previous model, the "Last Click" or "Last Interaction" model gives all the credit for a sale to the channel that makes it happen, regardless of the consumer's path.

In our previous examples, in

B2B: this time the website would account for 100% of the sale

and in B2C: also

It is useful to measure conversion and direct customers to the channel that maximizes the sale. It is only interesting if your conversion cycle is short, like for example if you have a crack on your windshield, you will be in a hurry and will be converted very quickly into a customer.

The disadvantage, as with the "First Click" model, is the singularity of the channel (the others not being integrated). This makes it difficult to have an overall view of the customer journey.

C - LAST INDIRECT CLICK

Here, the channel that converts is not taken into account. It is the second to last channel that accounts for 100% of the sale made. It is therefore the last channel on which the customer clicked before making a purchase that will be valued.

This model is consistent when you consider that inbound consumers with direct access already know about your company through other channels.  

It is only interesting in short cycle and allows you to value your acquisition campaigns. It is the default model of Google Analytics.

D - LAST ADWORDS CLICK

In this model, only the last Google Ads ad the customer clicked on gets 100% of the conversion. Even if the customer ends up coming back to the site and making a purchase directly on it, it is the Google Ads that will get the credit for the sale.

This is the model used if you want to highlight the investments made in Google Ads. It's the right model if the majority of your media investments are in Google Paid Search.

The main flaw of these 4 models is that they only focus on one channel. However, the customer journey has become more complex over time and we cannot give all the credit to a single channel. The prospect often goes through many phases before making a purchase, and therefore also through several contact points. Moreover, the channels do not necessarily play the same role and do not necessarily have the same impact. That's why I'm going to present you the 5 most used multi-touch models.

II - MULTI TOUCH MODELS

A - LINEAR

The linear model gives equal credit to all channels used by the consumer to make the purchase. It does not take into consideration the impact of each, all are given an equal share of the sale.

This model allows for a simple analysis of the channels involved in a purchase decision when multi-channel marketing campaigns are deployed.

The advantage is also to have a global view of the customer journey, allowing to improve and optimize it in its entirety.

On the other hand, optimization cannot be targeted, so it is difficult to determine the most effective touch points for conversion.    

B - DEPRECIATION OVER TIME

This model gives more and more credit to the channels that are close to conversion. It favors those that convert, over those that attract in the first place.

The advantage of this model is that it takes into account the entire customer journey and determines which channels had the greatest impact in closing the sale.

On the other hand, it is not the most suitable model if the time of reflection and research before the act of purchase is long.

This model makes sense in a B2B context where, for example, a prospect downloads a white paper, not specifically indicating a desire to buy, but where the multiplication of contact points afterwards (a phone call with a sales representative, then a physical meeting, etc.) will make the prospect evolve in his thinking and his desire to buy.

C - ASSESSMENT OVER TIME

This model is the exact opposite of the previous one. All touch points are included in the assessment of the sale, but those favored are those at the beginning of the journey.

This model is particularly interested in winning new prospects.

It is therefore not suitable if the company's goal is to optimize conversion, but it is perfectly suitable for acquisition or awareness research.  

D - POSITION BASED (U-SHAPED MODEL)

The position-based model, also known as the U-shaped model, gives importance to the first and last points of contact with the prospect. It is a mix between the "First Click" and "Last Click" model, while giving a little credit to the intermediate channels.

The first and last interactions each get 40% of the credit for the sale. The remaining 20% is divided evenly among the channels that make up the rest of the buying journey.  

Advertisers like this model because it adds value to their presence while giving credit to the channel that delivers the conversion.

The risk of this model is that important contact points are minimized or less important ones are over-valued.

E - DATA-BASED (ALGORITHMIC)

This last model is the one used to detail the behavior of prospects and consumers. At each touch point, data is collected and used to determine the weight of each channel in the final purchase decision. It takes into account the ad format and the time between an interaction with the ad and the conversion.

The advantage of this model is obviously the precision and accuracy of its analysis in relation to your business. It works thanks to intelligent algorithms and adjusts in real time to the evolution of your customers' journey thanks to "machine learning". It is therefore the closest to reality and the most dynamic. It allows you to identify the keywords, the ads, and the campaigns that are most likely to help you achieve your business objectives.

The disadvantage is that you can only use it if you have a sufficient and qualitative database. It is the most complex model to implement, a more mature marketing team will be required for this model. Moreover, Google does not give any explanation on how it values one channel over another in the purchase decision.

CONCLUSION

You also have the possibility to customize your attribution model without using an algorithm and by defining yourself the weight of each channel according to your objectives following criteria such as position, traffic level or others.

There is no perfect and universal model, each one has advantages and disadvantages. Moreover, you can choose different models depending on the target (new customer or existing customer). The model you choose must be adapted to your company, to your business model but also to your objectives. It depends on whether you want to analyze acquisition channels, conversion channels, customer journey details, etc. The factors of your business model come into play: Is the reflection phase long? Do you communicate on time-limited offers? Do you want to develop your brand awareness? These are all questions that will help you define the most suitable model for you.

You can also compare different models and choose the one that best suits your business model and why not even choose different ones at each stage of the journey to define what works and what doesn't (acquisition, retention, cross-selling and up-selling, all along the customer journey).

Also note that for a good ROI analysis, it is necessary to take into account the time spent between the first interaction and the conversion, but also to be aware that an advertisement seen but not clicked on by the prospect, can play a role in the conversion.

You should know that Google is planning to change its default model for all accounts in order to adapt to the evolutions in terms of privacy. This is the "data-driven" model that starts to be applied since October and will be progressively applied to all accounts by the beginning of 2022.

It will of course be possible to manually change this allocation model for the others we have discussed.

‍