As traditional, brick & mortar business has given way to online transactions, more commonly known as e-commerce, marketing practices have evolved as well. As marketers’ advertising spend transitioned into digital marketing formats, this widely expanded the data available, but also brought along enormous complexity.
The general goal behind marketing attribution is to identify the user actions that contributed, at least partially, to a desired outcome. These outcomes can range from an e-commerce transaction to a subscriber acquisition or a lead generation, depending on the type of business being marketed. In the digital space, each event per session can be tracked and analyzed to provide a clear picture of how an individual engaged with the available digital channels. Each engagement can be assigned a value and when combined with events from other individuals, this provides a relatively clear picture of the worth of a particular digital channel.
Part of the complexity with marketing attribution comes in which attribution model to use. While there are plenty of models available, they can generally be broken down into three large groups.
Single Source Attribution: Models in this group — for example the last-click model — attribute all value to a single source.
Fractional Attribution: These models attempt to attribute value to all of the events along the conversion path. This is usually done by weighting, such as the equal weight model, which divides value equally amongst all events.
Algorithmic Attribution: This type of attribution modelling uses data science to define an algorithm, usually proprietary, to determine the amount of credit for each channel.
In digital marketing attribution, there is no singular “correct” attribution model for everyone. Each business must assess their own digital channel, and at times, use different models to measure the effectiveness of all marketing channels.