Attribution Modeling: All you need to know for your Marketing Strategy

The dashboard of our Online Marketing platforms like Google Ads or Facebook Ads might give us the feeling that the conversions that we see are the conversions that we really get.
We might even be tempted to think that users simply click on one ad, spend a bit of time on our website and then simply decide to buy our product online.
But is that really what happens? Well, here are some numbers:

  • Almost 80% of all transaction value involves at least 2 marketing channel interactions.
  • 49% of the conversion value comes from transactions with 12 or more points in the conversion path.
  • Almost 98% of websites’ visitors will not make a purchase on their first visit.
  • 55% of the visitors to a website will leave during the first 15 seconds.
  • In online shops, the cart abandonment rate is close to 70%.

This highlights the hidden complexity of a conversion funnel: it takes time and several steps to convert a prospective client into an actual client and, in consequence, this process is difficult to track.

Question: what is the solution to this problem?

Answer: : attribution modeling!

What is Attribution Modeling

Attribution Modeling is the set of parameters that determine how the credits for a conversion are distributed to each touchpoint (or point of engagement) within a conversion path.

To make it crystal clear, let’s look at the following example.

A person is browsing Facebook, and he sees an ad for your product. He clicks on it and gets informed about your offer. A couple of days later, the same person searches for your company on Google and clicks on the organic result corresponding to your website but doesn’t buy anything. The next day, this person types in Google a search term that correlates with one of your products, clicks on your ad, and is then redirected to your product page and finally make a purchase.

We have here one conversion: a purchase, and three touchpoints: a click on a Facebook Ad, a click on an organic Google result and a click on a paid Google result.
Each of these touchpoints has played a role in the conversion. The purpose of attribution modeling is to distribute the most accurate “weight” to each of them.

Attribution modeling enables marketers to compare the efficiency of different marketing channels or marketing actions. It is an essential tool for anyone wishing to set-up an advanced data-driven marketing strategy.

What is the best attribution model?

First, let’s make 2 statements:

1. There is not a best attribution model

The most relevant attribution model depends on the type of conversion that the company is seeking. More precisely, it depends on the length of your conversion funnel.

If the conversion is easy such as a free app download that can be done with one touch of a smartphone screen, there will probably be a short amount of time between the moment the customer discovers the app and the moment he or she downloads the app.

If this is the case, it is better to choose an attribution model that gives more importance to the previous interactions before the conversion.

However, if the conversion is a big deal for the client for buying an expensive and complicated product (for example a car). There is no way the client will take any action quickly. He will gather and read technical specifications, customer and expert reviews, will think about everything for some time, and will then finally make a decision.

In this case, it makes sense to choose an attribution model that gives more weight to interactions that happened steadily before the conversion.

2. None of the existing attribution models are perfect.

We should not forget that the goal of an attribution model is first and foremost to give an approximation of the impact a marketing channel has on a conversion. Even if a marketer chooses the best attribution model for his business, the result will still be an approximation. But surprisingly, it is not a problem. A reliable approximation is all we need to make wise marketing decisions.

The 6 most popular attributions models

Last-click (or last touch): Although it isn’t the best in many cases, it has been the default attribution model for a long time. It might be a good choice for an easy (or effortless) conversion that doesn’t need a lot of touchpoints before happening. However, other options are better if the conversion needs effort and reasoning (example: an expensive product).

Last non-direct click: This is similar to the Last click attribution model, but with improvements. It excludes direct traffic. Indeed, if somebody types the website URL or bookmarks the website and then generates a conversion, the question we should ask is what channel made him enter or save the website address. This attribution model will give credit to the previous touchpoint.

First-click: This is the opposite of the Last-click attribution model. It gives all the credit to the beginning of the conversion path and, in consequence, maximizes conversions coming from new customers. This model is, in many cases, too unbalanced to be relevant.

Linear: This attribution model gives the same credit to every touchpoint. The advantage is that no touchpoint is left behind. The disadvantage, however, is that the first and last interactions might be undervalued while middle interactions might be overvalued.

U shaped: This attribution model gives most of the credit to the first and the last touchpoints (usually 40% to each, but many tools allow you choose the ratio) and shares the rest of the credit between the touchpoints in the middle. This attribution model is relevant in many cases because it spreads the credit relevantly between touchpoints. However, its relevance highly depends on the type of conversion and conversion funnel.

Time decay: This attribution model will give the highest credit to the last touchpoint and increasingly less to the other touchpoints as they move further from the conversions. This model is definitely worth considering, but depending on your business, might overvalue the end of the conversion path.

Attribution Modeling is an aspect of Online Marketing you might take in account if you want to be able to say that you truly utilize Data-Driven Marketing, and it might even prevent the overestimation or underestimation of some marketing channels. The first step is to understand your own business and define your own conversion path in order to choose the attribution model that is right for you.