By Yacov Salomon, Head of Data Science at Krux
Our Krux Data Science team is very fortunate. Each day we work with many of the world’s leading marketers and media companies as they combine their scale and reach with the scale and reach of the Krux platform to gather data a massive amount of rich data and important insights about the consumers they serve. This is not something we take for granted. Just a few years ago most data scientists would have begged to gain access to the amount of data that the Krux platform process in just a 24-hour period.
The advent of cloud computing, the global networking of millions of digital devices and the rapid “data-fication” of everything has altered that paradigm forever. Krux was built to help enterprises make use of this flood of data, at scale, and effectively respond to this paradigm shift. The result is that each month Krux sees more than 3.5 billion devices and browsers; processes 5 billion CRM records and sends more than 200 billion targeting instructions to a myriad of execution systems. All of this happens within a dynamic 20-petabyte environment.
This massive trove of data offers our clients hugely valuable consumer insights that can help them orchestrate more relevant and more meaningful media, content, and commerce experiences. The scale of this data, even at the individual client level, is massive and requires algorithm-driven machine-learning to separate signal from noise and identify patterns and insights.
Recently, the Krux Data Science team undertook a new research initiative, the result of which is “The Global State of Data for Marketers 2016”, based on an analysis of more than 100 billion data collection events. This report leverages Krux’s unique reach and visibility across the web, understanding how consumers engage with marketers and media companies across their digital devices. The source data was aggregated, anonymized, and normalized by Krux’s Data Science team, and the team’s analysis provides insights on global device usage, and looks at correlation between consumer engagement and factors like household income (HHI), age and gender. Further, it shares recommendations and best practices for handling huge data volumes and analyzing consumer trends, based on our experience in support of the efforts of some of the world’s largest marketers and media companies.
Some interesting findings from the report include:
- When looking across genders, some targeting may not always lead to more consumer action, at least when using ‘click-through’ as the measure. Male consumers are delivered the highest volume of targeted impressions, yet they generate fewer clicks, with lower aggregate click-through rates compared to female consumers.
- When looking at mobile usage across geographies, the divide between mobile and desktop continues to grow, but is not happening at the same pace everywhere. While the US and Canada consumers generate a nearly equal volume of page views from desktop and mobile devices, users in most of the rest of the world generate most of their page views from mobile devices.
- When looking at usage of mobile, in ‘in-app’ environments, the device type may influence depth of engagement. Page view volumes are the highest on Apple devices in Europe and Asia Pacific regions, despite the fact that Android devices dominate those markets.