Since the advent of online commerce, marketers are seeking solution to one unresolved question. How do you evaluate performance of online campaigns? Particularly, at a time when global organisations are increasingly relying on online advertising to improve visibility. It is true that measuring offline campaigns can be equally tricky, but online measurement is titillating. Simply because, there are enough mechanisms to employ a tangible measurement tool and marketers dwell with an increased expectation of being able to track and measure online campaigns.
Also, without effective measurement, it is a little challenging to convince senior management, those particularly averse to the new world of digital media. Lack of a good attribution model skews marketing contributions either by over-inflating or by under reporting the performance of a particular campaign.
Typically, an average conversion will have multiple touch points. For illustrative purposes, I have tried to capture five common online channels, which assists the conversion path. In the below example, I have taken paid search as the user’s first touch point, before he/she visits other channels such as display/display retargeting, affiliates, organic search, social media and finally converts by visiting the website directly. Now we are clearly aware that a user interacted with 5 different channels before converting, but what we don’t know is how much of an influence all these channels have on the user’s buying behaviour.
Old school attribution logic says that since the user visits the website to convert, this particular event should be fully attributed with the conversion, which clearly is the last click model. But does that mean that there is little need or necessity to run campaigns on different channels? This is where it begins to get clouded.
Marketers have experimented with different attribution models to identify the significance of a particular campaign. For example, from the above instance where a user travels through five different touch points starting from paid search before converting, we could attribute campaigns the in the following ways.
- First click: Where Paid Search as the first touch point claims full credit for the conversion
- Last Click: Since the user comes directly to the website, this model completely ignores the initial touch points and attributes the conversion to the direct visit.
- Distributed: This is one of the simplest of attribution models where each channel is equally attributed.
- Linear: This particular model attributes the value to a conversion based on recency of the user interaction in an ascending order, with the last touch point receiving the highest attribution.
- Weighted: Weighted attribution model focuses on weighing the campaigns based on their significance.
We have seen the different attribution models that can be applied to campaigns. But does that explain which model we should be using? Yes, I personally feel that weighted attribution is the way to go. However, one must realise that there are some subjective considerations while using this model.
In the above example, I have indicatively weighed Paid Search and Direct Visit with the highest attribution, while sharing a smaller percentage between the other touch points. This is a completely subjective option. Instead of weighing Paid Search higher, you may feel that for your particular campaigns, a display ad which has an online promotion initiates the traffic push. In such an instance, you may want to weigh a higher attribution to the display ad. Or you may feel that the one of your high converting keywords gains significant keyword positions on search engines, resulting in higher traffic, for which you may consider weighing Organic search with a higher attribution. You may even want to award a higher attribution to the recency of the interaction.
Weighted attribution certainly allows you to attribute the conversions effectively, but is it ‘the’ model for attribution? Well, it is debatable. However, one must understand that to fully comprehend the impact of different channels, it is imminent that we have to delve deeper. What the weighted attribution model offers is an improvement over other attribution models.
Every morning, I reluctantly pick a copy of Metro at Station on my way to work. Reluctantly because, I am keen to utilise the short journey time by reading magazine subscriptions which I always leave unfinished otherwise. But somehow, I end up reading a Metro at least twice a week. Off late I have noticed quite a few adverts on Metro by Mcdonalds and Costa Coffee trumpeting how superior their coffees are in comparison with the others. Costa, understood. It’s been a Coffee Company and it’s promotions are not surprising. But McDonald’s?
If past successes are anything to go by, then 
