Last click attribution is going the way of the dinosaur. It had its time on the planet, but eventually a big meteor plunged down and *poof* no more last click. What is the big cataclysmic event? Quite simply put, the need for data.
We live in a world where multi-channel and omni-channel are common place terms. A world where it takes multiple touch points and brand exposures to convert a prospect. A world where Google research tells us 3,000 consumers studied took 3,000 different journeys to complete their purchase.
Last click was easy. It was simple. Whatever ad was clicked most recently must have driven the purchase. It’s also the default tracking mechanism in most web analytics tools like Google Analytics and SiteCatalyst. But savvy marketers know its not just the last click that drove the sale. We know that consumers are interacting with multiple ads on multiple devices to complete the sale.
Why does it matter how sales are attributed?
As marketers, we are under constant pressure to maximize our revenue while managing tight budgets. In order to overcome these obstacles, we need to know what channels are working hardest. Last click only gives credit for the last ad a consumer clicked, ignoring their first exposure and anything in between. We wouldn’t understand what marketing campaigns work better for introducing our products, and what works best at closing the sale. Not understanding the complete picture can lead to lost revenue and nobody wants that.
RetailMeNot has made a business out of exploiting last click attribution. Priceonomics does a great job of explaining how RetailMeNot capitalizes on affiliate commissions by getting people to search and click on their ads for coupons. By simply clicking on their link, the consumer is setting a cookie and giving credit to RetailMeNot for the sale – when all they did was have a good SEO campaign for coupons and provided little value to the consumer.
With the last click attribution model, we would think that our affiliate channel is performing well and may pull spend from other channels to better support it. This would be a mistake, would likely lead to a decline in sales.
What are better attribution models?
There are several different attribution models out in the world today. Avinash Kaushik has a great blog post on attribution, if you want to read more. I’m only going to touch on a few that I think are the most useful and practical.
1. Linear Attribution
If you are looking to move from Last Click, Linear attribution is probably the most simplistic to understand. In this model, each touchpoint in the purchase process gets equal credit for the transaction. If there were 5 touchpoints, each would get 20% credit.
2. Position Based Attribution
In a Position Based attribution model, 35%-40% credit is assigned to each the first and last touch point, and the remaining 20%-30% is distributed evenly to other interactions in the middle. In this example, the first and last touch point get 35% credit each, and the middle three get 10% each.
3. Time Decay Attribution
What’s next for online attribution?
Fortunately, web analytics tools are catching up to the needs of marketers. Google Analytics now has attribution models you can play with and see how the revenue per channel and touch point change. They even have the ability to develop your own custom models and share it with others in the Solutions Gallery. Similarly, IBM has a solution in their marketing suite as well.
Convertro was created by an in house marketing group that developed a custom algorithm based on user behavior, who in turn took their product to market. It is used by Comcast and Digitas, amongst others.
VisualIQ is a very sleek and robust tool that I have used in the past. The interface is sleek and it allows you to see real-time attribution, and is what I would consider a top of the line tool. They have a large list of clients that includes American Express, GM, Disney and AT&T.
By utilizing newer attribution models, analysts and marketers can uncover new ways to optimize marketing spend through understanding how each interaction impacts the sale.
Have additional tools or models to discuss? Add to the comments section.