Incremental growth is the holy grail of retail marketing. In the early days of online sales, retailers who went digital saw huge gains from performance marketing channels like paid search, display, and Google Shopping. As the online market has matured and these marketing channels have become more saturated, finding incremental gains has become far more challenging. Today, retailers feel the pressure to spend more to remain competitive despite flattening growth.
Instead retailers should spend smarter. That means taking a fresh look at how they budget marketing dollars and attribute sales. The most popular method of attribution, last touch, attributes 100% of the sale to the last interaction a consumer had prior to purchasing. But this attribution model overvalues bottom of the funnel marketing activities while undervaluing top and mid-funnel activities. If retailers reassess their attribution model, they may uncover untapped marketing channels and new sales opportunities.
Enter Multi-Touch Attribution
A smarter way to value different marketing initiatives is through multi-touch attribution. A multi-touch attribution model weights every interaction that led to the sale. The model may value those interactions differently, depending on when they occurred in the customer journey or on what channel. The highest value interactions earn the greatest credit, giving retailers a clearer picture of how their different marketing initiatives are performing and how to budget in the future.
There is no single approach to multi-touch attribution, and many retailers customize their attribution model in some way. But there are five common multi-touch attribution models that retailers can use as a framework for their own solutions. Each model has unique benefits and limitations. Following are some of those models, as well as insights into whether these attribution models will be a fit for your business.
For more insight on multi-touch attribution, including a step-by-step implementation guide, be sure to check out Sidecar’s latest Point of View, “Marketing Attribution Needs to Catch Up to Retail’s Omnichannel World”. This resource will help retailers make the case to the C-suite and explain why changing attribution models is critical for growth. The Point of View also provides invaluable insights that will help retailers plan this major shift.
Linear
Linear, or even-weighted, attribution is perhaps the simplest multi-touch attribution model. It attributes each interaction leading up to the sale equally. If there are five interactions, for example, each touchpoint will receive 20% of the credit.
This attribution model gives retailers greater insight into the middle touches in the customer’s journey that help drive a sale. In particular, this model could be useful for retailers with a long sales cycle where every interaction is important. It is also relatively simple to implement because there is no need for retailers to determine which particular interaction is more valuable than the other.
On the other hand, it’s very unlikely that every touchpoint on a customer’s journey is driving the same amount of value. To truly determine the value of each touch, retailers will need to implement a more sophisticated attribution model.
Time Decay
The time decay attribution model gives more value to interactions that happen later in the customer journey. This model assumes that upper and mid-funnel browsing and research activities will occur earlier in the customer journey, while the bottom-funnel assessment activities occur at the end of the journey.
This model may be attractive to retailers who want to optimize their bottom-funnel marketing efforts. Perhaps they are confident that they are attracting the right kind of consumers to their website but are having trouble converting. A time decay model will show retailers what is working best in these final stages of the shopping journey and attribute more credit to those activities.
Depending on the retail vertical, early touches can still be incredibly influential. This model will not value those interactions as highly and may encourage retailers to concentrate the majority of their spend on mid- and bottom-funnel marketing efforts.
U-Shaped
Also known as position-based attribution, a U-shaped model gives the most amount of credit to the first and last touch, for example giving first and last touches 40% credit each. Then the remaining 20% of credit for a sale is split evenly between the middle touches.
This model makes sense for retailers who value the first and last touches of the customer’s journey the most but want to give credit to the middle touches that nurtured them towards a purchase. This model can help retailers distribute their spend more effectively to the marketing activities that provide the first impression a customer has of the brand as well as the interactions that ultimately lead to a purchase.
W-Shaped
W-shaped attribution is similar to U-shaped, but provides more credit to a middle touchpoint in the customer journey. It splits credit equally from the first touch, a middle touch where a consumer converts to a lead, and the last touch. For example a W-shaped attribution model may credit these three touches 30% each, and the remaining 10% is split evenly between any other interactions.
This approach may fit the needs of B2B retailers or other retailers with a long sales cycle. Because the key middle interaction, lead conversion, is heavily weighted, this attribution model will help retailers improve their nurture activities and spend more dollars on these activities.
Algorithmic
The most complex but also the most accurate attribution model is algorithmic attribution. An algorithmic attribution is customized to your business needs. Each channel and touchpoint is weighted with its own unique value.
Algorithmic attribution values the interaction itself as opposed to when it happened in the customer journey, which is what sets it apart from other attribution models.
Retailers can determine the different weights of interactions manually or using machine learning. If retailers use machine learning, the algorithmic model will constantly evolve as new data and customer interactions occur. A manual approach will still require retailers to update the weights of different touches as customer behavior evolves.
