6 Cross device attribution
Cross device attribution is attribution where a marketer is able to connect a customer journey. For true true cross device attribution a marketer does not completely depend on cookies to connect interactions with his or her digital assets and/or marketing campaign, but employs a third parties device graph to identify users as they switch between devices.
How a device graph is important for attribution?
According to some recent studies, more than ninety percent of internet users use more than one device to complete a task over time. Now-a-days, customer behavior has become increasingly cross-device. A customer may search for an item on their smartphone and place the order from their laptop. This has made it increasingly difficult for marketers to recognize their customers. One had no other option but to treat one customer as three or four distinct people.
So what’s the solution? It is none other than using an attribution solution with a solid deterministic device graph that allows for customer journeys to be connected across devices. Any attribution or automated bidding solutions that is unable to track users as they move across devices to complete a task is going to provide a sub optimal result.
Imagine you run a site that sell novels. You advertise it on the Guardian and Google Adwords. One of your customers clicks on your ad while using Guardian mobile app in office but doesn’t buy any book immediately. Later after reaching home, he researches your product through Google on his desktop. He clicks on one of your Adwords and immediately buys your product from his dektop.
Without cross device attribution, all the credit will go to the ad that the customer clicked on his desktop and you will not be able to see the full customer journey (the click on your ad while using the Guardian app).
Cross device marketer helps the marketer to recognize the same customer whether that interaction occurs on a desktop, smartphone or tablet.
Working of a device graph
A device graph works by linking devices together. It links an individual to all the devices he or she uses whether it is his desktop, laptop or smartphone. The more (login) data a device graph provider has, the greater the certainty is that that device graph can make accurate connections between devices and users and the more reliable the recommendations are that come using that specific device graph.
Different types of data matching methodologies
There are two types of data matching methodologies- deterministic and probabilistic.
Deterministic matching uses information such as log in data or hashed email address to recognize individuals; no matter whichever device they are using whether it is their laptop or smartphone. For example when a customer purchases something online and puts his information such as name, address etc, it is deterministic data. Then when he will work from another device with same information, the brand will be able to know who that person is on both devices.
On the other hand, probabilistic data uses things like IP addresses, location data and browsers to recognize individuals. Weather forecasting is common example of probabilistic data. If weather forecast is showing eighty percent chance of rain then it means that in the past when there were same conditions like this then it rained eighty percent of the time.
So which is better from both of these- deterministic or probabilistic? The best answer is to ask yourself that which method can give you positive market returns. Most companies claim to be deterministic but very few have the data to provide the reach that large advertisers need.
Understanding what data sets are used is important. And make sure you agree with the way data is collected and managed so you don’t find your ads remarketing to users across different devices that do not want to be exposed to your brand.