Data lakes have been around for nearly ten years, and innovative marketers are dipping their toes in the water by building their own or thinking about it. Kicking off a marketing data lake project requires a detailed plan and strong collaboration between teams, among other musts, as these savvy folks are finding. It helps to have someone who’s been there. That’s why we’re hosting a hangout on marketing data lakes, featuring five panelists who have either built their own or consulted for brands across industries in building their data lakes.
Erin Kelly, a practice area lead for Slalom’s information management & analytics group, is one of our panelists. A data lake veteran, she shares what they are, the types of companies considering them, and three key points to bear in mind when building a marketing data lake:
RampUp: Erin, would you mind sharing what you do at Slalom and what part data lakes play in your day-to-day thinking?
Erin: I lead our information management and analytics practice. In the Boston office, we go to market in three main offerings. One is helping companies with their cloud strategy and creating modern data architecture, which data lakes certainly fall into as a key component of helping organizations accelerate their use of data. We also do work around self-service analytics and data management—really, everything that’s required to get the right data to the right people at the right time—so we handle people, process, and technology from an enablement standpoint. We also have a data science capability. Machine learning and artificial intelligence are big areas of focus for us in that space.
Data is a key component that we’re seeing across all conversations. I think as data gets more accessible, there’s a greater need to get business users closer to the raw data where they can do the analysis and manipulation and really drive data-driven decision making at their organization.
RampUp: That all makes sense, but I’m sure it takes a while to get to that realization. Can you back up for our audience who may not be familiar with data lakes and how marketers in particular can start to have conversations about them at work? Can you describe what a marketing data lake is and how long they’ve been around?
Erin: At its most basic definition, a data lake is a storage repository. The data lake terminology was coined around 2010 as folks were starting to explore and break down how to take a variety of data into an area for analysis. For us, when we think about a data lake, we think of leveraging a flat architecture to store data versus a dimensional model that you would have in your data warehouse world. One of the most valuable functions of a data lake is storing data in its most detailed form to enable new types of analysis today and in the future.
A primary factor of why “data lake” has become a commonplace term over the last eight or nine years is the fact that storage is relatively cheap. Especially with the onset of cloud technologies, storage cost is not the prohibitor any more, and it really allows organizations to accelerate the use of a wide variety of data.
RampUp: It sounds like the economics have worked themselves out to a degree. From an organizational perspective, how do you know if you’re ready to build a marketing data lake?
Erin: One of the big things we see is there certainly needs to be a business need for driving a data lake initiative. That need could be greater access to a variety and volume of data for analytics or perhaps a need to drive strategic innovation and capabilities within an organization. This underlying business need is critical to having a lasting, maintained data lake.
Equally important to the business need is clear ownership and a champion in place to help drive the effort. Often, that champion sits outside of the business organization under the CIO. He or she may sit in an architecture group or a business intelligence and analytics group. A rule of thumb is the champion should be someone who is going to think about the people and process around the overall technology of the data lake.
RampUp: Can you share some of the business needs that have been identified by some of your clients and how they differ between industries?
Erin: One example that we have wrapping up in Boston is in the retail footwear space. The brand had significant visibility challenges into the performance in the supply chain operations, to the point where they didn’t have the flexibility and nimbleness required to shift inventory from one place to another, whether that’s a distributor or a country or region. They used the concept of the data lake to get a lot of varied information across their supply chain into one area where they could start to ask some of the more complex questions involving both sales and marketing types of metrics and data, as well as financial and operations data. That blend of perspectives is one of the big benefits that the data lake had for this organization.
We’re currently at another organization here in Boston in the medical device space where they’re using the data lake to drive their manufacturing quality process. They want to take data from their SAP-type systems and other quality systems and blend them using cloud technology and a data lake as a foundation. This will help them to better understand their manufacturing quality process and identify areas for optimization.
RampUp: In terms of these two clients specifically, it sounds like marketing was a component but not necessarily the primary driving force for the creation of the data lake.
Erin: For these two specifically, no, but I can give you another example. We have another footwear company here in Boston where there was a challenge from a marketing standpoint of understanding the customer. So, whether that’s in the B2B channel, in the loyalty programs, or in events that this company would run, disparate silos of information were out there, both in on-premises and cloud data warehouses. The data lake served as a foundation to start to blend a single view of the customer that both marketing and sales could use to drive increased personalization, touch points, interactions, and optimal buying behavior.
RampUp: That makes sense. It seems that for the other clients you mentioned first, marketers at those companies would find it useful to understand the supply chain. They could capitalize on things that are available seasonally or drive people to places where a hot item just got topped up, for example. You wouldn’t necessarily know that without being able to analyze supply chain information in a data lake.
Erin: Yes. I think that’s where it’s interesting. The innovation on the marketing side is around understanding the journey of that customer throughout the process—not just knowing how that customer was acquired, but really know how are they interacting with the products and services that a company provides. Often, that type of data is siloed into large systems like SAP or enterprise-resource-type systems, where it has been difficult to unlock that information. But by leveraging a data lake, there’s a lower level of detail that companies can then use for marketing and sales initiatives.
RampUp: You mentioned some of the job titles that are associated with building data lakes. Other than folks under the CIO or those involved in data architecture, are there are any other job titles that come to mind for people intimately involved with building marketing data lakes?
Erin: First off, business representation is critical when you’re thinking through a data lake. A data lake is only as good as the business use cases that it starts to drive and the value companies get out of using a data lake. Having that kind of executive level buy-in on the business side in the marketing organization is key. Often within the operations aspect of the marketing organization, there may be analysts or other folks who are intimately involved in gathering and analyzing data today. Those folks are critical to understanding consumption patterns and knowing the answers to question such as, what are the key questions that folks are going to want to ask against data in a data lake? How do they want to receive that information?
Your design starts to push from the data lake into the consumption standpoint, but certainly those use cases can help influence how a marketing data lake is organized and the kind of architectural patterns that can truly unlock the value.
RampUp: Given that data lakes have been around since 2010, it looks like we’re entering a new decade of data lakes. What do you think are three important things for people to have in mind when considering building a marketing data lake in the next year or two?
Erin: Start with a plan and the initiative. Where we see the most success in folks moving into data lakes is that it’s not just dumping data into a storage area for storage sake. It’s really thinking through what’s the value that you want to unlock there and creating a design pattern and approach of how you’re going to use, maintain, and innovate from that data lake technology. That’s one.
Secondly, security is a big topic that we see in every industry. I think there’s this fear of data access, data privacy, and really thinking of security as a policy. But it has to be more than that. Pulling in folks across the organization that would be involved in security compliance-level decisions is critical, early and often. You may have the most beautifully designed marketing data lake from the plan that you put in place, but if there are security challenges, credential challenges, the people who need to access data don’t have rights and you haven’t thought that through from a design standpoint, you’re certainly not going to get the benefit you’re looking for.
Thirdly, I think what we’re seeing a lot is leveraging agile-type methodology as well as DevOps for data. Look for opportunities to have continuous integration. Use infrastructure to get the most out of your data lake foundation and sustain that for the future.
Erin: Yes, absolutely.
RampUp: Is there anything else you wanted to add at this point? Feel free to save any nuggets for July 10th when we have our hangout.
Erin: A data lake can be quite transformative. Like any change at an organization, it requires some mindfulness around thinking through how you sustain a data lake going forward. Think through not only the technical components and the underpinnings from the infrastructure standpoint, but also how should people be accessing that.
That tends to drive a different level of data knowledge that’s required in the organization. So, where functional areas like marketing may have some siloed processes that they do offline with different metrics and terminology around data, a data lake starts to require more governance and more common language of data to really drive the value that I think is absolutely available and out there by leveraging the data lake.
RampUp: Right. So, this project, similar to the digital transformation projects of the early aughts, is going to require teams to come more together to break down their own silos, as well as their data silos.
Erin: Yes, exactly. Absolutely.
RampUp: Great, Erin. Thanks again for your time and we look forward to speaking with you.
Have questions about data lakes? We’ll answer them during our data lake hangout tomorrow, July 10, 2018, at 9 am PDT. Register here.