.svg)
In a previous post, we explored the meaning of marketing attribution with a short story to illustrate our definition. At the end of the article, we concluded that there are various ways to measure attribution. There are dozens of attribution models, however, some patterns are considered as classic models. They all have perks and limitations, so I leave it up to you to make your own idea of the perfect attribution models.
To make it easier to grasp the various scenarios, let’s recap the customer journey of Donna, the heroine of our previous article:
Do you remember? Donna first clicked on a paid ad on Google, then clicked on a Facebook post, later on, a link in a newsletter, then on a sponsored tweet, and finally visited the online store directly to purchase her dress. So how should we value each of these steps?
Exclusive attribution models
These models attribute 100% of the credit for a conversion to only one touchpoint. They are the most basic and the most common patterns.
Last-click attribution
This model attributes 100% of the credit to the last touchpoint before the conversion. It ignores all the other steps during the customer journey. This is the attribution model per default in many analytical tools and the pattern that most companies use to measure their online advertising campaigns. In our made-up story, this would be the direct search.
Last non-direct click
This model works like the previous one, excluding direct traffic. So 100% of the credit goes to the sponsored tweet (second social media step).
Last Google/Bing/Facebook/Twitter/etc. click
This model focuses on the last ad clicked before a conversion happened. These models are very dependent on one specific channel. All other touchpoints are ignored. Here, if we consider a “last AdWord click” approach, then the first touchpoint would get 100% of the credit (it’s the only AdWord ever clicked in this scenario). This model is useful if your job is focused on one specific channel (as a social media manager, you might be interested only in the last ad clicked on Facebook or Twitter for instance).
First-click interaction
This model gives 100% of the credit to the first touchpoint, no matter its nature. All other clicks between the first click and the conversion are ignored. Incidentally, the paid search step would be, again, the one in our story.
These exclusive attribution models all have the same constraint: they exclude most steps of the customer journey to focus solely on one touchpoint.
Multi-touch attribution models
These attribution models are more interesting: they consider the customer journey as a whole and they split the credit amongst the different touchpoints.
Linear attribution
This model takes into consideration every single touchpoint throughout the customer journey. However, they are evenly credited – the respective impact of each touchpoint is not assessed. In this case, each step gets 20% of the credit.
Time decay attribution
This attributive model grants more credit to the touchpoints closest to the conversion. In Donna’s case, the direct search might receive 30% of the credit while the paid search only receives 5%. The major drawback of this pattern is the almost negation of the awareness factor – usually the first touchpoint, an essential stepstone.
Position based attribution
Also called U-shaped attribution and this model grants 40% of the attribution to the first and last touch point, splitting the remaining 20% evenly among the other touchpoints. In our example, paid search and direct search receive 40% each, while the other steps get 6,67% each. Although possibly the most advanced attribution model, it still distributes fixed values. There is no appreciation of the impact of each step.
Algorithmic attribution
Algorithmic attribution is like the Rolls Royce of marketing attribution. Not only does it consider each step in the customer journey, it also attributes a customized value to each of them. It requires the intervention of data scientists to build and maintain algorithms that can appraise the importance of each touchpoint in the customer journey. The more information you want to gather, the more algorithms you need. For example, you may want to track a customer journey from someone who uses several devices. Or you may want to measure the influence of your TV spots on your online customers. With connected objects, you might even track offline advertising! Imagine Donna is walking by a connected scrolling advertising board with her phone in her bag. Advanced attribution could detect the proximity of both connected object and estimate that the ad on the board was part of her customer journey.
So what is the right attribution model then?
This is a tough question to answer. There is no right or wrong, it depends on how you want to measure your data. The most important thing is for you to be aware of which attribution model you use to understand the relevance of your advertising reports.
Clearly, algorithmic attribution is more accurate than the other types of attribution. It helps you get past the hindrance of cross-channel advertising and cross-device users. In other words, you can see how you lure customers to you, regardless of how many devices they use or which advertising channel they are exposed to. Even the most reluctant clients are under the radar.
The thing is, algorithmic attribution is a rather expensive service since it mostly targets companies with large advertising budgets. Furthermore, the creation and the maintenance of algorithms demand high costs in technology and personnel. In the end, you may very well end up spending hundreds of thousands a year to get your hands on the precious data! Another issue with tailor-made attribution models is the time until data starts rolling out. It typically takes a few weeks, but it can be months before anything tangible appears. A serious upfront investment for smaller companies. Finally, there are many companies offering customer-journey insights, however, some are subject to a conflict of interest. If company X works as one of your ad servers and reports on your global advertising activities, how can you know for sure that company X’s competitors won’t be prejudiced? You just can’t.
Affordable algorithmic attribution – a distant dream?
Start-ups and small-to-medium companies may not afford an algorithmic attribution service, however, their advertising and marketing activities may already be too spread out and too complex to rely on a free analytical solution, like Google Analytics or Bing’s Webmaster Tools.
That’s why we decided to offer an affordable and independent solution that delivers AI-powered insights on your customer journeys. With packages starting at very reasonable prices, you get within 24 hours the first results – independent from any online advertising tycoons.