Attribution Modelling

Attribution modeling is the practice of mapping touchpoints to monetarily relevant events within a customer journey, which is directly related to the return on investment of e-commerce websites, campaigns, or rankings in the organic search results. Attribution refers to the assignment of a value to specific interactions made by users or customers in different channels. Each interaction is to a certain extent responsible for the results of a digital asset and receives a corresponding numerical value, with which the results can be relatively accurately quantified afterwards.

The aim of attribution modeling is to analyze the role of individual touchpoints and channels for the customer journey. Customers come to a website through a variety of media such as ads, a hyperlink, or an article in the SERPs. Attribution modeling is designed to identify these different touchpoints and their causal role with regard to the ROI or conversion rate, thereby helping to weight the touchpoints and channels in order to be able to utilize multichannel marketing more efficiently and make more informed budget decisions.

General information[edit]

Attribution or assignment generally refers to the attribution of psychological effects and motivations to their causes for actions. The attribution theory was formulated by Fritz Heider at the end of the 1990s. It forms the theoretical foundation for application in marketing, as it is customary today. An important point was the fact that, for example, a impressions of ads cannot be quantified because their effect on an actual purchase decision was not measurable. In addition, models such as cost per click (CPC), first click or last click sometimes did not provide correct attributions because they cannot show clearly which click was the decisive factor for a purchase decision which also caused a problem for distributing commissions in affiliate marketing.

Attribution is meant solve these problems by examining the causal connections between individual channels and purchases made and defining them within a general framework. Marketers are supposed to be able to look at how different channels and campaigns influence purchase decisions to be able to utilize the marketing budget as effectively as possible. An inventory of all channels and touchpoints between medium and user is therefore paramount to attribution modeling. Since conversion paths can be very different and sometimes individual channels overlap, these digital assets and their relationships to each other are quite important. Thus, multichannel tracking or tagging of the channels is essential.[1]

How it works[edit]

Attribution modeling initially comprises several channels, which can be selected depending on marketing objectives:

There are generally interactions at different touchpoints before a purchase takes place. For example, a user sees an AdWords ad after entering a search term. He clicks the ad and is directed to a website where he signs up for a newsletter to receive a discount voucher. A few days later he returns to the website and makes a purchase after completing a product and price comparison.

Google Analytics has different attribution models in order to assign these channels and user interactions in and between these channels to a later purchase:

  • Last interaction: In this model, the last visit to the site would be considered decisive. Google recommends this feature for products or services that do not have a long decision-making process so that you can approach potential customers during the purchase process.
  • Last non-direct click: In this case, all direct accesses would not be included. Instead, the last indirect click, which in the example described above would be the newsletter subscription to obtain the voucher, would be decisive for the purchase. Google uses this model as the default model for reports without multi-channel funnel. It can serve as a comparison for other models.
  • Last AdWords click: In this model, a purchase would be associated to the last click on an AdWords ad. In the example, this is also the first interaction. It shows which AdWords ads have resulted in the most conversions.
  • First interaction: The first interaction with a channel would count here. In the example, this is the AdWords ad or paid search. This model is mainly recommended for raising awareness and branding campaigns. Certain keywords and channels can be assigned bonus values to determine which of them has increased popularity.
  • Linear attribution: In linear attribution, all channels or touchpoints are considered equal and are assigned percentage values, whereby all channels combined comprise 100%. In the example, the AdWords ad, the newsletter, and a direct visit to the website would each be considered a touchpoint and assigned a value of 33.3%. This model can be selected for campaigns that affect customer contact and public awareness, since all touchpoints are given equal values.
  • Time decay: In this model, the touchpoints are considered in relationship to time. The touchpoints which are closest to the conversion or sale will get the greatest value. In the example, the website and the newsletter would be the triggering touchpoints. The more a touchpoint is removed from a purchase, the lower its assigned value for that sale. In time-based campaigns, this model can be used to distinguish the time-limited campaign from other touchpoints.
  • Position-based: Position-based attribution assigns the highest value to the first and last interactions. The AdWords ad and the last visit to the website would be given greater value than the newsletter subscription. This model is recommended if the launch of a brand on the market and the clicks that led to conversions are the most important touchpoints.

There are of course numerous factors that would make one or the other model preferable. Google offers a tool that highlights the effects on the ROI for a comparison of different models. It is also possible to create a custom attribution model to take into account business-related estimates and factors.

Relevance to search engine optimization[edit]

The correct association of channels and touchpoints can have many advantages for marketers. Usually, customer journeys are anything but linear. A customer journey usually involves several channels, devices and sessions. Attribution modeling helps to map these different customer journeys and thereby make them quantifiable. Attribution can answer questions such as what channels, campaigns, and touchpoints generate or are responsible for the highest turnover.

Marketers are thus given a detailed picture of the effectiveness of their activities and can identify cross-connections between channels, for example to identify ROPO effects, second screen usage, and cross-channel activities. At the same time, attribution allows the optimization of the activities so that the effectiveness of existing assets can be further increased. For example, content marketing and content with added value that initially generate traffic and then increase the reach of the digital assets. Marketers receive much better data when attribution modeling accurately depicts the touchpoints than with monitoring models that only consider the last click.


  1. Multi-Channel Attribution Modeling: The Good, Bad and Ugly Models Accessed on 11/15/2016

Web Links[edit]