Who doesn’t like to get more out of what they’ve already got? Organizations already collect tons of data about consumer usage and transactions, yet many – if not most – are using only part of that data, or are leaving additional benefits on the table.
Data monetization is all about getting more bang for the org’s buck – and there’s a lot of bang to be had for companies that play it smart. So what does it look like to “play it smart?”
Randy Koch, ARM Insight CEO, and Tim Sloane, VP of Payments Innovation at Mercator Advisory Group, break it all down. Sloane shares his thoughts on data monetization as an industry, how different data types are being used to accelerate the pace of innovation in machine learning and AI, and how organizations are using data monetization to drive business value and new revenue streams.
“Data monetization is all about leveraging the data that you have through new channels,” said Sloane. “The trick to all of this, of course, is that much of that data includes Personal Identifiable Information (PII). So, one of the major challenges is, how can I enable my data to be analyzed by others while making sure that individual information is not leaked during that process. What intrigued me about ARM Insight was their solution to that problem and their ability to aggregate data from a range of financial institutions to make that data available to others with the ability to protect that data.”
Koch says that ARM Insight has spent the past 5+ years taking data feeds and helping monetize data for more than 1,000 financial institutions, retailers and other third parties. As such, ARM Insight is in a position now to share the lessons learned on how to monetize data in the safest and most secure fashion that minimizes risk. Koch outlines a road map that companies can take to safely and securely monetize their data.
Four Key Themes of Data Monetization Road Map
Does selling data sound kind of shady? It’s not – at least, not for companies that take the time to learn their way around the three types of data. However, according to Koch, this misperception is among the top mistakes that keep organizations from making the most of their data. Here’s what Koch says every company needs to know before tackling the data monetization monster:
- Perceived fear is higher than actual risk. Once an organization is educated on the complexities of compliance and regulatory requirements, monetizing data is nowhere near as scary as many people make it out to be.
- Not all data is created equal. Try to treat data types the same, and you’ll end up in trouble, because the compliance and regulatory structures, as well as data monetization opportunities, are specific to each one.
- Data management matters. A lot. Machine learning and artificial intelligence (ML, AI) can’t use messy data. Keep it in a clean, modern data platform, and ML can use it. Store it in outdated silos, and no intelligence – artificial or live – will stand a chance of finding what is needed.
- Make a three-pronged monetization move. There are three different channels where data can potentially be monetized. Understand and leverage more channels to maximize revenue and value while minimizing risk to the data itself (and the customers who provide it).
Let’s dig a little deeper into these road map themes.
Creating Three Types of Data Enables Safe and Secure Data Monetization
Making data analyzable by others while protecting individual information from leaks during the process is one of the major challenges of monetizing data – but it is not insurmountable, because there are three types of data that can be used in different ways and with different levels of risk.
- Raw data with PII: ARM Insight CEO Randy Koch of Portland walks into a Starbucks at 7:02 a.m. and pays $2.12 for a black coffee. Also included: Koch’s credit card number, address, and email, or maybe his date of birth, or – depending on the circumstances – his Social Security number. This data type has a high level of risk, but is also valuable, so you need to be careful with it.
- Anomymized data: A Portland man walks into a Starbucks at 7:02 a.m. and pays $2.12 for a black coffee. This data has been scrubbed of any information that would point back to Koch. He cannot be identified, and his sensitive information cannot be stolen. Yet even without these details, the data remains useful in many of the same ways that raw PII is useful – simply with much less risk.
- Synthetic data: A Portland man walks into a Starbucks at 8:03 a.m. and pays $2.13 for a black coffee. This is a fake data set created from the core data. The numbers are close enough, or statistically relevant, to use for most legitimate purposes – i.e., analytics – but they are synthetic, or falsified… making it impossible to reverse engineer the transaction back to either an anonymous person or an actual, identifiable customer. Synthetic data, of course, carries a very low level of risk around its use, while anonymized data is slightly riskier and raw PII is the riskiest. However, that doesn’t mean organizations can’t use or even monetize all three data types – it just means they must be intimately familiar with the compliance and regulatory standards around doing so, with a well-articulated privacy structure in place to mitigate those risks.
Once an organization understands the three data types, embraces the compliance and regulatory structures within each type, and demolishes any remaining data silos, now it has a clear path forward on how to monetize the data and how to use machine learning to better the company.
Data Monetization Opportunities
So you’ve got some data. Now ask: What could you (or someone else) be doing with that data that you’re not already doing? Consider internal, external, and Innovation (AI/ML) initiative possibilities.
Internally, data can be cleansed and then fed into a new machine learning system for a new purpose – say, analyzing the organization’s customers and its operations. The organization could then build analytics for unique customer insights or sell additional analytics products based on its customers data back to those core customers.
Externally, there are many ways that others might like to leverage this data. For example, say that a fast food chain wants to know where people shop an hour before and an hour after dining with them. The chain is just one example of a third party that could benefit from data that the organization has in its possession. Another example might be a mall that’s considering including this fast-food chain in its layout. If the organization sells data to one or both of these parties, it creates a new revenue stream.
Finally, Innovation or AI/ML initiatives represent opportunities for organizations to think outside the box and do something truly innovative, creating brand new products based off the data. Think self-driving cars, or healthcare data that enables providers to identify skin cancer before it becomes a bigger issue. Many financial institutions have gone down this path to create new innovations like chat bots, digital assistants, and payments fraud detection solutions.
Of course, one does not simply slap a price sticker on data and sell it. There are rules, particularly around personally identifiable information (PII). Customers’ PII must be protected at all costs and should never leave a trail that third parties could follow back to the originators of the data (the company’s own customers) unless those individuals have opted in to participate in that way.
Data Organization: Emptying silos into the lake for AI/ML Success
This last section deserves to be first on every organization’s priority list when it comes to monetizing data, especially if AI or ML initiatives are part of its data-driven vision.
Sloane says ML and AI aren’t just relevant in analytics-driven departments like payments fraud and Suspicious Activity Reports (SARs) management. These new capabilities are spread across the organization from HR and legal to managing third parties. If AI solutions are everywhere, then the data that feeds them must also be everywhere. Data must be accessible, clean, and updated in real time so that it can be used to train algorithms.
“I’ve been researching machine learning tools for about two years now and when I recognized that ARM Insight was creating synthetic data out of a data set, I was very impressed,” said Sloane. “I’d written about this about two years ago as an interesting capability, but I didn’t expect it to turn into meaningful products in such a short period of time.”
There was a time when it made sense to keep data in silos. It was nice and neat and clearly delineated, and it worked fine with legacy systems. Today, however, Koch says those siloes have become the single biggest challenge to moving forward. It’s time to empty those siloes and start thinking of data storage more like a lake.
“Way too many organizations make that mistake,” Koch warned. “They get so in love with machine learning that they’re not blocking and tackling. First clean the data and get it into a format that machine learning can use and then allow the machine to do its job.”
ARM Insight has developed a White paper that outlines the “Road map to Safe Data Monetization” and highlights many of the points in this article. To learn more and get a copy of the White paper visit www.arminsight.com
Subscribe to our podcast via: