Big Data is a key buzzword in the market these days, and for good reason. Big Data deployed in digital media campaigns provides deeper insights than traditional analytics, including stats about people’s purchases, how they browse online, and their personal interests.
Big Data is also essential for modeling campaign attribution – assigning credit to a particular marketing channel or action for a click or view. While crucial for attribution, marketers must know the benefits of big data as well as the challenges.
Benefits of Big Data for Attribution
Big Data provides the information that underlies attribution models, especially multi-touch models. It helps marketers see the whole customer purchase journey from first touch to last and which channels or campaigns are working throughout the marketing funnel.
Big Data also helps marketers see what’s not working. Not all touch points have the same value or contribute equally to campaign goals and ROI. Marketers can leverage performance data to replace channels, tweak messaging, or pivot campaigns.
Big Data also helps marketers see the different considerations consumers make before performing an action. This more detailed views of customers allows marketers to tweak channels or messaging based on customer insights and decision-making.
In a more specific example, Big Data helps reconcile mobile device identity to let marketers see the entire conversion trail. This is currently done via two methods:
- Deterministic Identification uses a user’s known information to match devices. Most of these only work within one’s own property or sites (i.e. Facebook or logins), or with self-identifying data, such as email addresses.
- The Probabilistic Approach uses publicly available information for ad serving to assess the statistical probability and match devices, which is somewhere between 50-90% accurate.
Big Data Challenges
While Big Data provides a comprehensive picture of the consumer throughout the marketing funnel, it also comes with its own issues. Chief among them is how to gain control of all the data being provided in a timely fashion. Even when marketers have the capacity to analyze Big Data, they often find their access to it hampered. Not only can it be hard to capture data from some channels, but data collection and storage present their own set of challenges:
Markets typically use 5-10 data sources with separate license holders, departments, etc. Not only is data spread out, it is also often not readily available. Marketers may have to wait up to 6-9 months to gain access to the data they need.
After gaining access to the requested data, it must all be aligned, including online and offline information. DMPs organize and store data, but cannot run models. DMPs are also relatively new, which means that marketers, advertisers and agencies are still figuring out how to effectively use them.
By the time the organized information is sent to be modeled, a full year may have passed, meaning that all that real-time data collection does not translate into real-time actionable analysis.
The goal is to have more real-time insights into data and campaigns. Processing times aren’t there yet, but Big Data analysis is moving in that direction.
Takeaways for Marketers
Marketers are beginning to realize that post-campaign reporting isn’t enough. They want insights to directly influence the success of the campaign long before it ends. While that’s the eventual goal, it is not yet possible. According to eMarketer, 43% of markets say they have a solution in place, but there are gaps that still need to be addressed. Until that time, marketers will have to keep creatively using post-campaign insights to inform future campaigns.
photo: data center journal