How’s it going with figuring out how you’re going to use “big data” to get more return on your campaigns? Or understand which customers to target for upsells or new products? Or increase overall usability of your mobile and web channels?
If you’re like a lot of those who I speak to, you’re not quite sure what “big data” really is, let alone how it’s going to be harnessed and used for amping up your marketing programs. You’re certainly not alone. A recent InformationWeek survey found that only nine percent of respondents rated their organizations at extremely effective at identifying critical data and using it to make decisions.
How can you be successful at getting value from big data?
Your guiding principle is “Small is Beautiful" -- transforming big data into small data.
1. Break up big data into manageable data sets
Think of all of the marketing data that potentially comprises a big data marketing analytics set. Without breaking a sweat, I can think of nearly 20 sets. Each set is comprised of dozens of variables and dimensions:
- Visitor and visit behavioral data from fixed and mobile web
- Demographic and behavioral data from social media
- Visitor acquisition from partners, ads, search, social media
- Broadcast media view rates
- Response activity from emails and postal mail
- Voice of customer and survey data collected both online and offline
- Call center activity
- Offline activity, such as leads, sales, event interactions
- Customer lifecycle data, such as purchase or conversion history, revenue history
How many data sets are you working with now? One, two? Maybe web analytics and SEO and email …maybe PPC.
What you can do -- Start small by adding a few key data sets to your analytics framework and then maximizing the value of findings you can realize through faster queries available through big data analytics engines. Once you have digested the value of these starter data sets, then it's time to build on what you’ve learned.
2. Start with a simple question and build better segments through additional data and testing
I guess this sounds more like a riddle than a direction.
You want to start with a basic question that can be answered both through the data sets that you are beginning with and those you’d like to add. For example:
How do I identify prospects who will purchase our new service?
If you’re starting with web analytics and email analytics data, you could determine that one segment is more valuable than another segment:
- Visitors who respond to a specific email offer promotional price and visit the site more than twice within a monthly time frame from New York have a higher conversion rate than those from Florida.
You could test this segment and then add additional data, such as customer purchasing history, to show that visitors who have been customers for at least five years have higher conversion rates.
What you can do -- The analytics results from the big data are powerful when used to develop increasingly targeted segments that can be tested through marketing campaigns, customer experience analysis and lifetime customer value models.
3. Break apart the analysis while breaking down data silos
Managing and analyzing multiple data sets leads to the data silos we see in many organizations. It’s hard to control the data. That’s why it’s common for data to be analyzed in individual sets and not seen or used in its entirety. (More on this in my recent columns, The Digital Analytics Center of Excellence Dream Team and Digital Analytics Roadmap to Excellence).
Adding data to the big data infrastructure provides an opportunity to bring data owners together to answer the questions that may no longer be wedded to only one data set. Cooperative analysis is a somewhat unfamiliar concept in many organizations. However, I see this occurring today in organizations that will bring together web analytics, surveys, search engine referral and social media data to triangulate results and come up with recommendations on site optimization.
What you can do -- Form teams of data owners to work with their own and others’ data in the formation of questions, application of new data sets, development of tests and recommendations based on the big data analytics results.
Big data -- for all of the excitement it is generating -- is likely a few years away from generating the power it is capable of for many organizations. In the meantime, I think there is a lot of good habits marketers can practice using the data sets already available. Focusing on segmentation rather than broad data, conducting analysis to answer questions and then layering additional data on to the questions and seeking to bring more people into the analysis process are all the right types of foundational pieces that will set up your organization for big data analysis when it comes to a data warehouse or Mapreduce server cluster near you.
Title image courtesy of Alexsandar Mijatovik (Shutterstock)
Editor's Note: Read more of Phil's Big Data advice in How to Untangle the Data Deluge