Product Product | June 17, 2016

Cross-Device Segmentation: Mapping the User Purchase Journey Across Devices

The Internet has evolved, and so have your consumers. While earlier, the online shopping landscape consisted of tech-savvy consumers making purchases using desktop computers on various websites, today purchases are predominantly made on mobile devices. A recent report from Google indicates that 90% of consumers use multiple devices to accomplish online activities, such as researching and shopping online.

And with online shoppers owning multiple devices, the user journey has become more complex than ever before.  A user might browse for a product on his laptop at work, research about it on his smartphone on his way home, and make a purchase on his tablet at home. This makes it increasingly difficult to track the customer’s online behaviour.

The traditional audience tracking methodologies have left advertisers with broken links of the purchase journey and no means of putting the pieces together for a single customer. Therefore,  it has become imperative to devise new and advanced strategies to join the pieces together for the complete picture of the user journey, and thereby make informed marketing decisions.

Cross-device segmentation allows advertisers to do just this and helps them engage with the consumer better. Rupali Patel, Data Scientist, and Rohit Agarwal, Software Engineer at Media iQ throw more light on Media iQ cross-device segmentation solution.

Understanding cross-device segmentation

Rohit Agarwal
Rohit Agarwal, Software Engineer, Media iQ

Advertisers have always been interested in targeting audiences who showcase similar attributes or have similar purchase patterns. Until recently, categorising audiences based on their online activity was limited to their browser cookies and/or mobile device IDs of the browsers/devices that exhibited the activities of interest.

So, for example, if a consumer browsed through the travel pages of an online magazine website on his phone in the morning, travel advertisers would only be able to reach him on that same device, even if he spent the next few hours accessing the Internet via his work laptop.

However, this single-device approach is no longer applicable, as according to a recent report by Nielsen, the average American user is accessing content using an average of four different devices.

“To map the user journey on different devices, we have partnered with leaders in the cross-device data space to provide advertisers with the ability to understand their customers better, and enable them to achieve unprecedented levels of brand engagement. Media iQ’s current cross-device offering allows advertisers to add segments catering to cross-device users in their existing campaigns; tracks the performance of those segments, and helps them make informed decisions about their marketing strategies”, says Rohit.

Rupali further adds, “to run campaigns efficiently, advertisers need to come up with new strategies which will help them understand their consumers in depth – what are they searching for, where are they located, how much time do they stay on a particular page, etc. Since consumers today, exhibit an omnichannel shopping behaviour, different campaigns need to set their strategies differently.”

At Media iQ, we come across such scenarios where the aim is to target audiences in near real time across channels to achieve that specific campaign goal. For example, sometimes brands would like to re-target audiences, who are already actively searching or showing purchase intent on desktops, on their mobile or tablets.

Implementing cross-device segmentation

Rupali says, “Recently, we wanted to target users at an offline store on their mobile devices, and then re-target them on their desktops at a later stage. We were able to do this with the help of Media iQ’s hyperlocal product- Locate, where we found out their location and mobile device IDs. Then by using Adbrain’s device graph, we extracted desktop cookies associated with those devices, and targeted users across multiple screens.”

rupali patel
Rupali Patel, Data Scientist, Media iQ

So how it is done? Rohit adds, “The data provided by our data partner- Adbrain has to be made available to both batch and real-time services, so we needed a solution that would allow us to do that. Media iQ handles a peak traffic of around a million requests per second, and the servers are configured in such a way that it is able to extract interesting traits of the users from the request parameters, and tries to segment users based on the logic configured by the analysts. The interesting traits are represented in the form of user segment associations that are generated by the servers (the servers can generate associations up to 3 billion per day). These associations have to be synced with AppNexus servers to enable us to target the users based on the exhibited traits.”

The cross-device solution, therefore, needs to fit in between the generation of the associations and its sync with AppNexus, so that it can transparently resolve the cross-device IDs and sync them as well. It also needs to enable the analysts to target users across devices for custom segments, which may have been generated by hive or other systems using various other data sources.

The solution needs to provide high throughput, should be horizontally scalable and fault tolerant. A typical query to store data would be key value lookups, but as relational databases are not optimised for such queries, we needed the NOSQL database to expose the Adbrain data. We evaluated certain open source databases like Redis, Cassandra, Aerospike etc., and zeroed in on Aerospike as it fulfilled all the essential requirements mentioned above.

Aerospike also provides the option to store data in SSDs (Solid State Drives) with the performance comparable to what some in-memory data store would provide. Since Aerospike provides a cost-efficient SSD storage option, we could store the identified user traits against the user ID across time to make a historical user profile available in real time. The cross-device service can work in tandem with the recently introduced AppNexus instant audience service, allowing us to target users across devices in near real-time.

Media iQ’s cross-device solution enables analysts to specify the kind of audience segments that can be targeted across screens in near real time. It also allows them to track the performance of those segments and enables brands to make informed decisions about their marketing strategies.         

Connecting various touch points

There is no doubt that cross-device segmentation is critical for success today, as consumers use multiple technologies or touch points in their journey to the final purchase. And as cross-device targeting becomes more and more commonplace in campaigns, marketers need to educate themselves and understand the value of their data and serve contextually relevant content. What is also important is to better understand, and precisely measure, the value of all these devices and the roles they play in driving conversions.



Leave a Reply

Your email address will not be published. Required fields are marked *