Digital attribution has come into the spotlight owing to an increase in digital ad spend and the emphasis on the consumer journey. As the volume of digital data increases, so does the need for more accurate methods of measuring performance and/or ROI, moving the focus to more data-driven modelling approaches.
The continuously developing digital landscape makes it imperative for marketers to evolve their measurement techniques accordingly.
What is digital attribution? Digital attribution is the science of assigning credit from a conversion (sales, web visits, registrations, leads, etc.) to the marketing touchpoints that drove it. This, in turn, determines the marketing activities that are delivering true value for an advertiser.
The standard approach used for digital attribution is known as the “last-click model”, which awards 100 percent of the credit to the last click preceding the conversion. One of the primary benefits of this approach is that it allows advertisers to run readily available reports with little difficulty in a timely manner. However, there are obvious flaws with such an approach, out of which two key concerns are:
- It ignores any digital touchpoints prior to the last click
- It can overstate the true impact of “search”, as it frequently appears as the last touchpoint
Standard digital attribution approaches
The flawed last-click approach prompted digital planners to adopt other popular approaches. Based on our evaluation, we saw some inherent flaws in each approach, so we decided to create our own.
Our approach to digital attribution
Our analytical modeling process attributes ‘weights’ to various marketing touchpoints for their contribution toward driving a desired result. This approach looks beyond the last click and identifies awareness drivers and convertors in the consumer’s path. Combining exposures with site traffic data, we can analyze a user’s path to purchase and formulate recommendations. Eventually, by evaluating all touchpoints, we can see the role each digital channel plays in driving a conversion. The channel weights are derived by performing an uplift analysis as illustrated.
Calculating the weights
Calculating the conversion rates for two channels as shown in the figure above, we can see that Channel 1 has a weight of 0.5 (1%/2%) which in turn also gives Channel 2 a weight of 0.5 (sum of weights is equal to 1). This weighting is applied when Channel 1 and Channel 2 are both in the consumers’ path to conversion. The methodology is applied for each channel against all other channels used for the campaign, which then gives us weights to apply for each.
This approach allows us to uncover the true consumer journey and in turn allows us to fairly allocate the credit, not just among digital media channels, but also at a publisher level.
Key learnings
Here are some of our learnings based on three studies conducted across clients from various industries, media channels and publishers in the MENA region:
- Search activity continues to perform well, but not as well as shown in the last-click approach.
- Within “social”, the performance by platform varies greatly, depending on the campaign objective and industry vertical.
- The performance of certain channels and publishers can vary significantly depending on the KPI of the campaign, so, it’s always recommended that each campaign be measured on its own merit.
- Pacing, i.e., managing investments over the campaign period with publishers, rather than sudden bursts of activity, has been proven to show significant uplifts in results.
Attribution modelling has enabled planning teams to optimize campaigns and focus investment on those channels and publishers that are driving conversions. This has resulted in a step change in performance and this approach will increasingly become the de facto way to analyze digital campaign performance.
