With the anticipated growth of real-time bidding and audience buying in 2011, brands will be able to manage a much larger volume of online advertising.
A few years ago, marketers discovered that most of the conversions happened without an associated click in display advertising. Then, a little while later, view-based conversions were added to the performance metric. While it was a small step forward, the obvious question is: if the last click is not always indicative of the user's decision to convert, why would the last view be any better? In short, it's not.
In an earlier ClickZ column, I discussed the importance of going beyond the last click or last ad impression and described why an incorrect attribution model will lead to suboptimal results. Well, a new year calls for a new model - multi-touch attribution where all interactions within an association window are considered to have influences over a user's conversion.
The power of digital advertising offers more than just targeting with higher precision; it also provides instantaneous feedback on how ads perform. With consumers exposed to increasingly more ad impressions and interactions of different media types, it's become harder than it should be to explain how each event along the path to conversion affects the user's decision. In the analytics sense, the current flawed last-touch attribution model suffers from the lack of a probability framework. Since the last touch (click or view) is defined by the conversion event, it's difficult to calculate the probability. Therefore, it's nearly impossible to measure the true influence of the last touch.
Although it's challenging to reconstruct the causation model that drives each consumer's decisions and responses, we can use a probability model to approximate users' behaviors statistically. Essentially, this inference model is similar to the predictive model. In this case, every important factor of the user interactions is encoded into input variables and the decision the user made is the outcome variable. Each factor contributes to the outcome by raising or lowering the probability of the user conversion. Think of this as similar to conducting dozens of A/B tests on the influence of each factor within the decision process.
For any given ad campaign, this type of attribution model can have two primary use cases. The first is retroactive reporting, which is the basic use case of an attribution model. The goal is to compare the contribution of each type of touch point/consumer event. The second use case is proactive analysis. This new use case mirrors a "what-if" analysis where the user in each ad call can be scored based on previous behavior and touch points. Given the user's past interactions with the brand through ad impressions, clicks, and site visits, we can further infer what type of interaction may incur the highest lift of the probability of user response.
For instance, let's say the multi-touch attribution model has determined that users who have seen three different types of ad creative for a new car in the last three days have a 10 percent higher probability to convert than users who have only seen two different types of ad creative for the same car. For a user who has seen two types of ad creative in the last two days, we would choose one of the new creatives and bid to the value of having an incremental 10 percent gain in conversion probability.
The example here indicates a future change from traditional bidding algorithms in two ways.
With the anticipated exponential growth of real-time bidding and audience buying in 2011, brands will be able to manage a much larger volume of online advertising. Scale drives efficiency, while efficiency enables more optimization. So let me conclude by attempting at a "Moore's Law" for online advertising: for every doubling of volume, the transaction efficiency of advertising will increase 50 percent; for every doubling of transaction efficiency, the complexity of bidding algorithm will increase 50 percent.
As Chief Technology Officer at Turn, Xuhui Shao focuses on the power of optimization, machine learning, and advanced analytics solutions in driving new business models, products, and services across all industries. Xuhui is responsible for architecting the machine learning and optimization technology to deliver the most effective data-driven digital advertising in the world. He is passionate about the dynamic online advertising community and works closely with industry leaders developing data transparency and consumer privacy protection.
For the last 12 years, Xuhui has practiced research and development in machine learning, statistical theory, and computational intelligence for Fortune 100 companies in various industries from banking, finance, online retailing, healthcare, insurance, marketing, and online advertising. As the lead inventor and co-inventor of three awarded patents in the areas of advanced analytics and optimization, Xuhui is a recognized expert in harnessing data and transforming analytics into actionable insights and optimization strategies.
He earned his bachelor's and master's of science degrees from Tsinghua University, Beijing, and his Ph.D. in electrical engineering from the University of Minnesota.
US Consumer Device Preference Report
Traditionally desktops have shown to convert better than mobile devices however, 2015 might be a tipping point for mobile conversions! Download this report to find why mobile users are more important then ever.
E-Commerce Customer Lifecycle
Have you ever wondered what factors influence online spending or why shoppers abandon their cart? This data-rich infogram offers actionable insight into creating a more seamless online shopping experience across the multiple devices consumers are using.
September 9, 2015
12pm ET/9am PT
September 16, 2015
12pm ET/9am PT
September 23, 2015
12pm ET/ 9am PT